Papers

The cost of developing new drugs is estimated at billions of dollars per year. Identification of new molecules for drugs involves scanning existing bio-medical literature for relevant information. As the potential drug molecule is novel, retrieval of relevant information using a simple direct search is less likely to be productive. Identifying relevant papers is therefore a more complex and challenging task, which requires searching for information on molecules with similar characteristics to the novel drug. In this paper, we present the novel task of ranking documents based on novel molecule queries. Given a chemical molecular structure, we wish to rank medical papers that will contribute to a researcher's understanding of the novel molecule drug potential. We present a set of ranking algorithms and molecular embeddings to address the task. An extensive evaluation of the algorithms is performed over the molecular embeddings, studying their performance on a benchmark retrieval corpus, which we share with the community. Additionally, we introduce a heterogeneous edge-labeled graph embedding approach to address the molecule ranking task. Our evaluation shows that the proposed embedding model can significantly improve molecule ranking methods. The system is currently deployed in a targeted drug delivery and personalized medicine research laboratory.

poster session 4 @ red 6, poster session 9 @ red 6, poster session 4, poster session 9

Uncertainty estimation is an essential step in the evaluation of the robustness for deep learning models in computer vision, especially when applied in risk-sensitive areas. However, most state-of-the-art deep learning models either fail to obtain uncertainty estimation or need significant modification (e.g., formulating a proper Bayesian treatment) to obtain it. Most previous methods are not able to take an arbitrary model off the shelf and generate uncertainty estimation without retraining or redesigning it. To address this gap, we perform a systematic exploration into training-free uncertainty estimation for dense regression, an unrecognized yet important problem, and provide a theoretical construction justifying such estimations. We propose three simple and scalable methods to analyze the variance of outputs from a trained network under tolerable perturbations: infer-transformation, infer-noise, and infer-dropout. They operate solely during the inference, without the need to re-train, re-design, or fine-tune the models, as typically required by state-of-the-art uncertainty estimation methods. Surprisingly, even without involving such perturbations in training, our methods produce comparable or even better uncertainty estimation when compared to training-required state-of-the-art methods.

poster session 4 @ red 4, poster session 11 @ red 4, poster session 4, poster session 11

Deep neural networks are vulnerable to input deformations in the form of vector fields of pixel displacements and to other parameterized geometric deformations \eg translations, rotations, etc. Current input deformation certification methods either (\textbf{i}) do not scale to deep networks on large input datasets, or (\textbf{ii}) can only certify a specific class of deformations, \eg only rotations. We reformulate certification in randomized smoothing setting for both general vector field and parameterized deformations and propose \textsc{DeformRS-VF} and \textsc{DeformRS-Par}, respectively. Our new formulation scales to large networks on large input datasets. For instance, \textsc{DeformRS-Par} certifies rich deformations, covering translations, rotations, scaling, affine deformations, and other visually aligned deformations such as ones parameterized by Discrete-Cosine-Transform basis. Extensive experiments on MNIST, CIFAR10, and ImageNet show competitive performance of \textsc{DeformRS-Par} achieving a certified accuracy of $39\%$ against perturbed rotations in the set $[-10\degree,10\degree]$ on ImageNet.

poster session 4 @ red 2, poster session 11 @ red 2, oral session 4 @ red 2, poster session 4, poster session 11, oral session 4

Weakly supervised temporal sentence grounding aims to temporally localize the target segment corresponding to a given natural language query, where it provides video-query pairs without temporal annotations during training. Most existing methods use the fused visual-linguistic feature to reconstruct the query, where the least reconstruction error determines the target segment. This work introduces a novel approach that explores the inter-contrast between videos in a composed video by selecting components from two different videos and fusing them into a single video. Such a straightforward yet effective composition strategy provides the temporal annotations at multiple composed positions, resulting in numerous videos with temporal ground-truths for training the temporal sentence grounding task. A transformer framework is introduced with multi-tasks training to learn a compact but efficient visual-linguistic space. The experimental results on the public Charades-STA and ActivityNet-Caption dataset demonstrate the effectiveness of the proposed method, where our approach achieves comparable performance over the state-of-the-art weakly-supervised baselines. The code is available at https://github.com/PPjmchen/Composition_WSTG.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

Adversarial training is so far the most effective strategy in defending against adversarial examples. However, it suffers from high computational costs due to the iterative adversarial attacks in each training step. Recent studies show that it is possible to achieve fast Adversarial Training by performing a single-step attack with random initialization. However, such an approach still lags behind state-of-the-art adversarial training algorithms on both stability and model robustness. In this work, we develop a new understanding towards Fast Adversarial Training, by viewing random initialization as performing randomized smoothing for better optimization of the inner maximization problem. Following this new perspective, we also propose a new initialization strategy, backward smoothing, to further improve the stability and model robustness over single-step robust training methods. Experiments on multiple benchmarks demonstrate that our method achieves similar model robustness as the original TRADES method while using much less training time (~3x improvement with the same training schedule).

poster session 4 @ red 2, poster session 11 @ red 2, poster session 4, poster session 11

In many games, moves consist of several decisions made by the player. These decisions can be viewed as separate moves, which is already a common practice in multi-action games for efficiency reasons. Such division of a player move into a sequence of simpler / lower level moves is called splitting. So far, split moves have been applied only in forementioned straightforward cases, and furthermore, there was almost no study revealing its impact on agents' playing strength. Taking the knowledge-free perspective, we aim to answer how to effectively use split moves within Monte-Carlo Tree Search (MCTS) and what is the practical impact of split design on agents' strength. This paper proposes a generalization of MCTS that works with arbitrarily split moves. We design several variations of the algorithm and try to measure the impact of split moves separately on efficiency, quality of MCTS, simulations, and action-based heuristics. The tests are carried out on a set of board games and performed using the Regular Boardgames General Game Playing formalism, where split strategies of different granularity can be automatically derived based on an abstract description of the game. The results give an overview of the behavior of agents using split design in different ways. We conclude that split design can be greatly beneficial for single- as well as multi-action games.

poster session 4 @ blue 6, poster session 9 @ blue 6, poster session 4, poster session 9

We propose JoTA (Job Taxonomy Alignment), a domain-independent, knowledge-poor method for automatic taxonomy alignment of lexical taxonomies via word embeddings. JoTA associates all the leaf terms of the origin taxonomy to one or many concepts in the destination one, employing a scoring function, which merges the score of a hierarchical method and the score of a classification task. JoTA is developed in the context of an EU Grant aiming at bridging the national taxonomies of EU countries towards the European Skills, Competences, Qualifications and Occupations taxonomy (ESCO) through AI. The method reaches a 0.8 accuracy on recommending top-5 occupations and a wMRR of 0.72.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

Multiagent reinforcement learning algorithms have not been widely adopted in large scale environments with many agents as they often scale poorly with the number of agents. Using mean field theory to aggregate agents has been proposed as a solution to this problem. However, almost all previous methods in this area make a strong assumption of a centralized system where all the agents in the environment learn the same policy and are effectively indistinguishable from each other. In this paper, we relax this assumption about indistinguishable agents and propose a new mean field system known as Decentralized Mean Field Games, where each agent can be quite different from others. All agents learn independent policies in a decentralized fashion, based on their local observations. We define a theoretical solution concept for this system and provide a fixed point guarantee for a Q-learning based algorithm in this system. A practical consequence of our approach is that we can address a `chicken-and-egg' problem in empirical mean field reinforcement learning algorithms. Further, we provide Q-learning and actor-critic algorithms that use the decentralized mean field learning approach and give stronger performances compared to common baselines in this area. In our setting, agents do not need to be clones of each other and learn in a fully decentralized fashion. Hence, for the first time, we show the application of mean field learning methods in fully competitive environments, large-scale continuous action space environments, and other environments with heterogeneous agents. Importantly, we also apply the mean field method in a ride-sharing problem using a real-world dataset. We propose a decentralized solution to this problem, which is more practical than existing centralized training methods.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

We propose a novel zero-shot multi-frame image restoration method for removing unwanted obstruction elements (such as rains, snow, and moire patterns) that vary in successive frames. It has three stages: transformer pre-training, zero-shot restoration, and hard patch refinement. Using the pre-trained transformers, our model is able to tell the motion difference between the true image information and the obstructing elements. For zero-shot image restoration, we design a novel model, termed SiamTrans, which is constructed by Siamese transformers, encoders, and decoders. Each transformer has a temporal attention layer and several self-attention layers, to capture both temporal and spatial information of multiple frames. Only unsupervisedly pre-trained on the denoising task, SiamTrans is tested on three different low-level vision tasks (deraining, demoireing, and desnowing). Compared with related methods, SiamTrans achieves the best performances, even outperforming those with supervised learning.

poster session 4 @ red 3, poster session 8 @ red 3, oral session 8 @ red 3, poster session 4, poster session 8, oral session 8

Probabilistic learning to rank (LTR) has been the dominating approach for optimizing the ranking metric, but cannot maximize long-term rewards. Reinforcement learning models have been proposed to maximize user long-term rewards by formulating the recommendation as a sequential decision-making problem, but could only achieve inferior accuracy compared to LTR counterparts, primarily due to the lack of online interactions and the characteristics of ranking. In this paper, we propose a new off-policy value ranking (VR) algorithm that can simultaneously maximize user long-term rewards and optimize the ranking metric offline for improved sample efficiency in a unified Expectation-Maximization (EM) framework. We theoretically and empirically show that the EM process guides the leaned policy to enjoy the benefit of integration of the future reward and ranking metric, and learn without any online interactions. Extensive offline and online experiments demonstrate the effectiveness of our methods.

poster session 4 @ blue 2, poster session 7 @ blue 2, poster session 4, poster session 7

We report a new neural backdoor attack, named Hibernated Backdoor, which is stealthy, aggressive and devastating. The backdoor is planted in a hibernated mode to avoid being detected. Once deployed and fine-tuned on end-devices, the hibernated backdoor turns into the active state that can be exploited by the attacker. To the best of our knowledge, this is the first hibernated neural backdoor attack. It is achieved by maximizing the mutual information (MI) between the gradients of regular and malicious data on the model. We introduce a practical algorithm to achieve MI maximization to effectively plant the hibernated backdoor. To evade adaptive defenses, we further develop a targeted hibernated backdoor, which can only be activated by specific data samples and thus achieves a higher degree of stealthiness. We show the hibernated backdoor is robust and cannot be removed by existing backdoor removal schemes. It has been fully tested on four datasets with two neural network architectures, compared to five existing backdoor attacks, and evaluated using seven backdoor detection schemes. The experiments demonstrate the effectiveness of the hibernated backdoor attack under various settings.

poster session 4 @ red 2, poster session 11 @ red 2, oral session 4 @ red 2, poster session 4, poster session 11, oral session 4

Evaluating interactive AI (IAI) systems is a challenging task, as their output highly depends on the performed user actions. As a result, developers often depend on limited and mostly qualitative data derived from user testing to improve their systems. In this paper, we present InteractEva; a systematic evaluation framework for IAI systems. InteractEva employs (a) a user simulation backend to test the system against different use cases and user interactions at scale with (b) an interactive frontend allowing developers to perform important quantitative evaluation tasks, including acquiring a performance overview, performing error analysis, and conducting what-if studies. The framework has supported the evaluation and improvement of an industrial IAI text extraction system, results of which will be presented during our demonstration.

poster session 4 @ blue 1, poster session 11 @ blue 1, poster session 4, poster session 11

Program translation aims to translate source code from one programming language to another. It is particularly useful in applications such as multiple-platform adaptation and legacy code migration. Traditional rule-based program translation methods usually rely on meticulous manual rule-crafting, which is costly both in terms of time and effort. Recently, neural network based methods have been developed to address this problem. However, the absence of high-quality parallel code data is one of the main bottlenecks which impedes the development of program translation models. In this paper, we introduce CoST, a new multilingual Code Snippet Translation dataset that contains parallel data from 7 commonly used programming languages. The dataset is parallel at the level of code snippets, which provides much more fine-grained alignments between different languages than the existing translation datasets. We also propose a new program translation model that leverages cross-lingual snippet denoising auto-encoding and Multilingual Snippet Translation (MuST) pre-training. Extensive experiments shows that the multilingual snippet training is effective in improving program translation performance, especially for low-resource languages. Moreover, our training method shows good generalizability and consistently improves the translation performance of a number of baseline models. The proposed model outperforms the baselines on both snippet-level and program-level translation, and achieves state-of-the-art performance on CodeXGLUE translation task.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

A significant challenge in credit risk models for underwriting is the presence of bias in model training data. When most credit risk models are built using only applicants who had been funded for credit, such non-random sampling predominantly influenced by credit policymakers and previous loan performances may introduce sampling bias to the models, and thus alter their prediction of default on loan repayment when screening applications from prospective borrowers. In this paper, we propose a novel data augmentation method that aims to identify and pseudo-label parts of the historically declined loan applications to mitigate sampling bias in the training data. We also introduce a new measure to assess the performance from the business perspective, loan application approval rates at various loan default rate levels. Our proposed methods were compared to the original supervised learning model and the traditional sampling issue remedy techniques in the industry. The experiment and early production results from deployed model show that self-training method with calibrated probability as data augmentation selection criteria improved the ability of credit scoring to differentiate default loan applications and, more importantly, can increase loan approval rate up to 8.8\%, while keeping similar default rate comparing to baselines. The results demonstrate practical implications on how future underwriting model development processes should follow.

poster session 4 @ red 6, poster session 8 @ red 6, poster session 4, poster session 8

In 2020, the White House released the “Call to Action to the Tech Community on New Machine Readable COVID-19 Dataset,” wherein artificial intelligence experts are asked to collect data and develop text mining techniques that can help the science community answer high-priority scientific questions related to COVID-19. The Allen Institute for AI and collaborators announced the availability of a rapidly growing open dataset of publications, the COVID-19 Open Research Dataset (CORD-19). As the pace of research accelerates, biomedical scientists struggle to stay current. To expedite their investigations, scientists leverage hypothesis generation systems, which can automatically inspect published papers to discover novel implicit connections. We present automated general purpose hypothesis generation systems AGATHA-C and AGATHA-GP for COVID-19 research. The systems are based on the graph mining and transformer models. The systems are massively validated using retrospective information rediscovery and proactive analysis involving human-in-the-loop expert analysis. Both systems achieve high-quality predictions across domains in fast computational time and are released to the broad scientific community to accelerate biomedical research. In addition, by performing the domain expert curated study, we show that the systems are able to discover ongoing research findings such as the relationship between COVID-19 and oxytocin hormone. All code, details, and pre-trained models are available at https://github.com/IlyaTyagin/AGATHA-C-GP.

poster session 4 @ red 6, poster session 9 @ red 6, poster session 4, poster session 9

Learning effective embeddings for potentially irregularly sampled time-series, evolving at different times scales, is fundamental for machine learning tasks such as classification and clustering. Task-dependent embeddings rely on similarities between data samples to learn effective geometries. However, many popular time-series similarity measures are not valid distance metrics, and as a result they do not reliably capture the intricate relationships between the multi-variate time-series data samples for learning effective embeddings. One of the primary ways to formulate an accurate distance metric is by forming distance estimates via Monte-Carlo-based expectation evaluations. However, the high-dimensionality of the underlying distribution, and the inability to sample from it, pose significant challenges. To this end, we develop an Importance Sampling based distance metric -- I-SEA -- which enjoys the properties of a metric while consistently achieving superior performance for machine learning tasks such as classification and representation learning. I-SEA leverages Importance Sampling and Non-parametric Density Estimation to adaptively estimate distances, enabling implicit estimation from the underlying high-dimensional distribution, resulting in improved accuracy and reduced variance. We theoretically establish the properties of I-SEA and demonstrate its capabilities via experimental evaluations on real-world healthcare datasets.

poster session 4 @ red 4, poster session 11 @ red 4, poster session 4, poster session 11

This paper studies a basic function fitting problem, which we coin \textit{first-order convex fitting} (FCF): given any two vector sequences $\{\bvec{x}_i\}_{i\in[T]}$ and $\{\bvec{p}_i\}_{i\in[T]}$ in $\mathbb{R}^d$, when is it possible to \emph{efficiently} construct a convex function $f(\bvec{x})$ that ``fits'' the two sequences in the first-order sense, i.e, its gradient $\nabla f(\bvec{x}_i)$ equals $\bvec{p}_i$ for all $i \in [T] = \{ 1,\cdots, T \}$? Despite the close connection between function fitting and machine learning, FCF has surprisingly been overlooked in the past literature. With an efficient constructive proof, we discover a clean answer to this question: \name{} is possible \textit{if and only if} the two sequences are \textit{permutation stable}: $\sum_{i=1}^{T} \bvec{x}_i \cdot \bvec{p}_i \geq \sum_{i=1}^T \bvec{x}_i \cdot \bvec{p}_{\sigma(i)}$ for any permutation $\sigma$ of $[T]$. We demonstrate the usefulness of FCF in two applications. First, we empirically show how it can be used as a surrogate to significantly accelerate the minimization of the original convex function. Second, we study how it can be used as an empirical risk minimization procedure to learn the original convex function. We provide efficient PAC-learnability bounds for special classes of convex functions learned via FCF, and demonstrate its application to multiple economic problems where only function gradients (as opposed to function values) can be observed.

poster session 4 @ red 4, poster session 11 @ red 4, oral session 4 @ red 4, poster session 4, poster session 11, oral session 4

Ensemble Learning is an effective method for improving generalization in machine learning. However, as state-of-the-art neural networks grow larger, the computational cost associated with training several independent networks becomes expensive. We introduce a fast, low-cost method for creating diverse ensembles of neural networks without needing to train multiple models from scratch. We do this by first training a single parent network. We then create child networks by cloning the parent and dramatically pruning the parameters of each child to create an ensemble of members with unique and diverse topologies. We then briefly train each child network for a small number of epochs, which now converge significantly faster when compared to training from scratch. We explore various ways to maximize diversity in the child networks, including the use of anti-random pruning and one-cycle tuning. This diversity enables "Prune and Tune" ensembles to achieve results that are competitive with traditional ensembles at a fraction of the training cost. We benchmark our approach against state of the art low-cost ensemble methods and display marked improvement in both accuracy and uncertainty estimation on CIFAR-10 and CIFAR-100.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

Although deep learning has made significant progress on fixed large-scale datasets, it typically encounters challenges regarding improperly detecting unknown/unseen classes in the open-world scenario, over-parametrized, and overfitting small samples. Since biological systems can overcome the above difficulties very well, individuals inherit an innate gene from collective creatures that have evolved over hundreds of millions of years and then learn new skills through few examples. Inspired by this, we propose a practical collective-individual paradigm where an evolution (expandable) network is trained on sequential tasks and then recognize unknown classes in real-world. Moreover, the learngene, i.e., the gene for learning initialization rules of the target model, is proposed to inherit the meta-knowledge from the collective model and reconstruct a lightweight individual model on the target task. Particularly, a novel criterion is proposed to discover learngene in the collective model, according to the gradient information. Finally, the individual model is trained only with few samples on the target learning tasks. We demonstrate the effectiveness of our approach in an extensive empirical study and theoretical analysis.

poster session 4 @ blue 2, poster session 7 @ blue 2, oral session 7 @ blue 2, poster session 4, poster session 7, oral session 7

Utterance rewriting aims to recover coreferences and omitted information from the latest turn of a multi-turn dialogue. Recently, methods that tag rather than linearly generate sequences have proven stronger in both in- and out-of-domain rewriting settings. This is due to a tagger's smaller search space as it can only copy tokens from the dialogue context. However, these methods may suffer from low coverage when phrases that must be added to a source utterance cannot be covered by a single context span. This can occur in languages like English that introduce tokens such as prepositions into the rewrite for grammaticality. We propose a hierarchical context tagger (HCT) that mitigates this issue by predicting slotted rules (e.g., "besides _") whose slots are later filled with context spans. HCT (i) tags the source string with token-level edit actions and slotted rules and (ii) fills in the resulting rule slots with spans from the dialogue context. This rule tagging allows HCT to add out-of-context tokens and multiple spans at once; we further cluster the rules to truncate the long tail of the rule distribution. Experiments on several benchmarks show that HCT can outperform state-of-the-art rewriting systems by up to 17.8 BLEU points.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

Vulnerability to adversarial attacks is a well-known weakness of Deep Neural networks. While most of the studies focus on single-task neural networks with computer vision datasets, very little research has considered complex multi-task models that are common in real applications. In this paper, we evaluate the design choices that impact the robustness of multi-task deep learning networks. We provide evidence that blindly adding auxiliary tasks, or weighing the tasks provides a false sense of robustness. Thereby, we tone down the claim made by previous research and study the different factors which may affect robustness. In particular, we show that the choice of the task to incorporate in the loss function are important factors that can be leveraged to yield more robust models.

poster session 4 @ blue 6, poster session 9 @ blue 6, oral session 9 @ blue 6, poster session 4, poster session 9, oral session 9

As an independent social and economic entity, game servers plays a dominant role in building a stable, living, and attractive virtual world in massive multi-player online role-playing games (MMORPGs). We propose and implement a novel intelligent decision support system for server merge (SM) for maintaining the game ecology at the macro level. The services provided by this system include server health diagnosis, server merge assessment, and combination strategy recommendation. Specifically, we design an effective time series prediction algorithm to diagnose the health status of one server (e.g., user activity, online time, daily revenue) based on real game scenarios, and then select the servers with poor status from all servers. Moreover, to dig out the inherent development laws of servers from the historical merge records, we leverage a correlation measurement algorithm to find the historical merged servers that are similar to the servers to be merged and then evaluate the potential trend after merging, which can assist experts to make reasonable decisions. We deploy our system into practice for multiple MMORPGs and achieve sound online performance endorsed by the game operation team.

poster session 4 @ blue 1, poster session 11 @ blue 1, poster session 4, poster session 11

In bi-objective search, we are given a graph in which each directed arc is associated with a pair of non-negative weights, and the objective is to find the Pareto-optimal solution set. Unfortunately, in many practical settings, this set is too large, and therefore its computation is very time-consuming. In addition, even though bi-objective search algorithms generate the Pareto set incrementally, they do so exhaustively. This means that early during search the solution set covers is not diverse, being concentrated in a small region of the solution set. To address this issue, we present a new approach to subset approximation of the solution set, that can be used as the basis for an anytime bi-objective search algorithm. Our approach transforms the given task into a target bi-objective search task using two real parameters. For each particular parameter setting, the solutions to the target task is a subset of the solution set of the original task. Depending on the parameters used, the solution set of the target task may be computed very quickly. This allows us to obtain, in challenging road map benchmarks, a rich variety of solutions in times that may be orders of magnitude smaller than the time needed to compute the solution set. We show that by running the algorithm with an appropriate sequence of parameters, we obtain a growing sequence of solutions that converges to the full solution set. We prove that our approach is correct and that Bi-Objective A* prunes at least as many nodes when run over the target task.

poster session 4 @ blue 4, poster session 8 @ blue 4, oral session 4 @ blue 4, poster session 4, poster session 8, oral session 4

Graph matching aims to establish correspondences between vertices of graphs such that both the node and edge attributes agree. Various learning-based methods were recently proposed for finding correspondences between image key points based on deep graph matching formulations. While these approaches mainly focus on learning node and edge attributes, they completely ignore the 3D geometry of the underlying 3D objects depicted in the 2D images. We fill this gap by proposing a trainable framework that takes advantage of graph neural networks for learning a deformable 3D geometry model from inhomogeneous image collections, i.e. a set of images that depict different instances of objects from the same category. Experimentally we demonstrate that our method outperforms recent learning-based approaches for graph matching considering both accuracy and cycle-consistency error, while we in addition obtain the underlying 3D geometry of the objects depicted in the 2D images.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

Generalized structural equations models (GSEMs) [Peters and Halpern 2021], are, as the name suggests, a generalization of structural equations models (SEMs). They can deal with (among other things) infinitely many variables with infinite ranges, which is critical for capturing dynamical systems. We provide a sound and complete axiomatization of causal reasoning in GSEMs that is an extension of the sound and complete axiomatization provided by Halpern [2000] for SEMs. Considering GSEMs helps clarify what properties Halpern's axioms capture.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

Zero/few-shot transfer to unseen services is a critical challenge in task-oriented dialogue research. The Schema-Guided Dialogue (SGD) dataset introduced a paradigm for enabling models to support an unlimited number of services without additional data collection or re-training through the use of schemas. Schemas describe APIs in natural language, which models consume to understand the services they need to support. However, the impact of the choice of language in these schemas on model performance remains unexplored. We address this by releasing SGD-X, a benchmark for measuring the robustness of dialogue systems to linguistic variations in schemas. SGD-X extends the SGD dataset with crowdsourced variants for every schema, where variants are semantically similar yet stylistically diverse. We evaluate two top-performing dialogue state tracking models on SGD-X and observe that neither generalizes well across schema variants, measured by joint goal accuracy and a novel metric for measuring schema sensitivity. Finally, we present a simple model-agnostic data augmentation method to improve schema robustness and zero-shot generalization to unseen services.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

Understanding how the predictions of deep learning models are formed during the training process is crucial to improve model performance and fix model defects, especially when we need to investigate nontrivial training strategies such as active learning, and track the root cause of unexpected training results such as performance degeneration. In this work, we propose a time-travelling visual solution DeepVisualInsight (DVI), aiming to manifest the spatio-temporal causality while training a deep learning image classifier. The spatio-temporal causality demonstrates how the gradient-descent algorithm and various training data sampling techniques can influence and reshape the layout of learnt input representation and the classification boundaries in consecutive epochs. Such causality allows us to observe and analyze the whole learning process in the visible low dimensional space. Technically, we propose four spatial and temporal properties and design our visualization solution to satisfy them. These properties preserve the most important information when (inverse-)projecting input samples between the visible low-dimensional and the invisible high-dimensional space, for causal analyses. Our extensive experiments show that, comparing to baseline approaches, we achieve the best visualization performance regarding the spatial/temporal properties and visualization efficiency. Moreover, our case study shows that our visual solution can well reflect the characteristics of various training scenarios, showing good potential of DVI as a debugging tool for analyzing deep learning training processes.

poster session 4 @ blue 2, poster session 7 @ blue 2, oral session 4 @ blue 2, poster session 4, poster session 7, oral session 4

The ability to study "gain of function" mutations has important implications for identifying and mitigating risks to public health and national security associated with viral infections. Numerous respiratory viruses of concern have RNA genomes (e.g., SARS and flu). These RNA genomes fold into complex structures that perform several critical functions for viruses. However, our ability to predict the functional consequence of mutations in RNA structures continues to limit our ability to predict gain of function mutations caused by altered or novel RNA structures. Biological research in this area is also limited by the considerable risk of direct experimental work with viruses. Here we used small functional RNA molecules (ribozymes) as a model system of RNA structure and function. We used combinatorial DNA synthesis to generate all of the possible individual and pairs of mutations and used high-throughput sequencing to evaluate the functional consequence of each single- and double-mutant sequence. We used this data to train a machine learning model (Long Short-Term Memory). This model was also used to predict the function of sequences found in the genomes of mammals with three mutations, which were not in our training set. We found a strong prediction correlation in all of our experiments.

poster session 4 @ blue 2, poster session 7 @ blue 2, poster session 4, poster session 7

Explainably estimating confidence in published scholarly work offers opportunity for faster and more robust scientific progress. We develop a synthetic prediction market to assess the credibility of published claims in the social and behavioral sciences literature. We demonstrate our system and detail our findings using a collection of known replication projects. We suggest that this work lays the foundation for a research agenda that creatively uses AI for peer review.

poster session 4 @ blue 1, poster session 11 @ blue 1, poster session 4, poster session 11

Missing value imputation is crucial for real-world data science workflows. Imputation is harder in the online setting, as it requires the imputation method itself to be able to evolve over time. For practical applications, imputation algorithms should produce imputations that match the true data distribution, handle data of mixed types, including ordinal, boolean, and continuous variables, and scale to large datasets. In this work we develop a new online imputation algorithm for mixed data using the Gaussian copula. The online Gaussian copula model produces meets all the desiderata: its imputations match the data distribution even for mixed data, improve over its offline counterpart on the accuracy when the streaming data has a changing distribution, and on the speed (up to an order of magnitude) especially on large scale datasets. By fitting the copula model to online data, we also provide a new method to detect change points in the multivariate dependence structure for mixed data with missing values. Experimental results on synthetic and real world data validate the performance of the proposed methods.

poster session 4 @ red 4, poster session 11 @ red 4, oral session 4 @ red 4, poster session 4, poster session 11, oral session 4

Demosaicing is the crucial step in the image processing pipeline and is a highly ill-posed inverse problem. Recently, various deep learning based demosaicing methods have achieved promising performance, but they often design the same nonlinear mapping function for different spatial location and are not well consider the difference of mosaic pattern for each color. In this paper, we propose a deep spatial adaptive network (SANet) for real image demosaicing, which can adaptively learn the nonlinear mapping function for different locations. The weights of spatial adaptive convolution layer are generated by the pattern information in the receptive filed. Besides, we collect a paired real demosaicing dataset to train and evaluate the deep network, which can make the learned demosaicing network more practical in the real world. The experimental results show that our SANet outperforms the state-of-the-art methods under both comprehensive quantitative metrics and perceptive quality in both noiseless and noisy cases.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

Recently, data-driven inertial navigation approaches have demonstrated their capability of using well-trained neural networks to obtain accurate position estimates from inertial measurement units (IMUs) measurements. In this paper, we propose a novel robust Contextual Transformer-based network for Inertial Navigation (CTIN) to accurately predict velocity and trajectory. To this end, we first design a ResNet-based encoder enhanced by local and global multi-head self-attention to capture spatial contextual information from IMU measurements. Then we fuse these spatial representations with temporal knowledge by leveraging multi-head attention in the Transformer decoder. Finally, multi-task learning with uncertainty reduction is leveraged to improve learning efficiency and prediction accuracy of velocity and trajectory. Through extensive experiments over a wide range of inertial datasets (e.g., RIDI, OxIOD, RoNIN, IDOL, and our own), CTIN is very robust and outperforms state-of-the-art models.

poster session 4 @ red 1, poster session 7 @ red 1, oral session 4 @ red 1, poster session 4, poster session 7, oral session 4

Posters are broadly used to present the important points of academic papers and can be seen as a special form of document summarization. However, the problem of automatic poster generation is under-investigated. In this paper, we present PosterBot, an automatic poster generation system for academic papers. Given a scholarly paper, PosterBot takes three steps to generate the poster. It first selects the most important sections, and then generates corresponding panels from them. Finally, all panels are integrated to get the complete poster. The demonstration shows the efficacy of our proposed system.

poster session 4 @ blue 1, poster session 11 @ blue 1, poster session 4, poster session 11

To interpret uncertainty estimates from differentiable probabilistic models, recent work has proposed generating a single Counterfactual Latent Uncertainty Explanation (CLUE) for a given data point where the model is uncertain. We broaden the exploration to examine δ-CLUE, the set of potential CLUEs within a δ ball of the original input in latent space. We study the diversity of such sets and find that many CLUEs are redundant; as such, we propose DIVerse CLUE (∇-CLUE), a set of CLUEs which each propose a distinct explanation as to how one can decrease the uncertainty associated with an input. We then further propose GLobal AMortised CLUE (GLAM-CLUE), a distinct, novel method which learns amortised mappings that apply to specific groups of uncertain inputs, taking them and efficiently transforming them in a single function call into inputs for which a model will be certain. Our experiments show that δ-CLUE, ∇-CLUE, and GLAM-CLUE all address shortcomings of CLUE and provide beneficial explanations of uncertainty estimates to practitioners.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

This paper tackles the problem of semi-supervised learning when the set of labeled samples is limited to a small number of images per class, typically less than 10, problem that we refer to as barely-supervised learning. We analyze in depth the behavior of a state-of-the-art semi-supervised method, FixMatch, which relies on a weakly-augmented version of an image to obtain supervision signal for a more strongly-augmented version. We show that it frequently fails in barely-supervised scenarios, due to a lack of training signal when no pseudo-label can be predicted with high confidence. We propose a method to leverage self-supervised methods that provides training signal in the absence of confident pseudo-labels. We then propose two methods to refine the pseudo-label selection process which lead to further improvements.The first one relies on a per-sample history of the model predictions, akin to a voting scheme. The second iteratively up-dates class-dependent confidence thresholds to better explore classes that are under-represented in the pseudo-labels. Our experiments show that our approach performs significantly better on STL-10 in the barely-supervised regime,e.g. with 4 or 8 labeled images per class.

poster session 4 @ blue 6, poster session 9 @ blue 6, poster session 4, poster session 9

We revisit the planning problem in the blocks world, and we implement a known heuristic for this task. Importantly, our implementation is biologically plausible, in the sense that it is carried out exclusively through the spiking of stylized neurons. Even though much has been accomplished in the blocks world over the past five decades, we believe that this is the first algorithm of its kind. The input is a sequence of symbols encoding an initial set of block stacks as well as a target set, and the output is a sequence of motion commands such as ``put the top block in stack 1 on the table''. The program is written in the Assembly Calculus, a recently proposed computational framework meant to model computation in the brain by bridging the gap between neural activity and cognitive function. Its elementary objects are assemblies of neurons (stable sets of firing neurons whose simultaneous firing is believed to signify the subject thinking of an object, concept, word, etc.), its commands include project and merge, and its execution model is based on widely accepted tenets of neuroscience. A program in this framework essentially sets up a dynamical system of neurons and synapses that eventually, with high probability, accomplishes the task. The purpose of this work is to establish empirically that reasonably large programs in the Assembly Calculus can execute correctly and reliably; and that rather realistic --- if idealized --- higher cognitive functions, such as planning in the blocks world, can be implemented successfully by such programs.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

Recurrent neural networks have proven effective in modeling sequential user feedbacks for recommender systems. However, they usually focus solely on item relevance and fail to effectively explore diverse items for users, therefore harming the system performance in the long run. To address this problem, we propose a new type of recurrent neural networks, dubbed recurrent exploration networks (REN), to jointly perform representation learning and effective exploration in the latent space. REN tries to balance relevance and exploration while taking into account the uncertainty in the representations. Our theoretical analysis shows that REN can preserve the rate-optimal sublinear regret even when there exists uncertainty in the learned representations. Our empirical study demonstrates that REN can achieve satisfactory long-term rewards on both synthetic and real-world recommendation datasets, outperforming state-of-the-art models.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

Tensor optimization is crucial to massive machine learning and signal processing tasks. In this paper, we consider tensor optimization with a convex and well-conditioned objective function and reformulate it into a nonconvex optimization using the Burer-Monteiro type parameterization. We analyze the local convergence of applying vanilla gradient descent to the factored formulation and establish a local regularity condition under mild assumptions. We also provide a linear convergence analysis of the gradient descent algorithm started in a neighborhood of the true tensor factors. Complementary to the local analysis, this work also characterizes the global geometry of the best rank-one tensor approximation problem and demonstrates that for orthogonally decomposable tensors the problem has no spurious local minima and all saddle points are strict except for the one at zero which is a third-order saddle point.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

Ontology-based query answering with existential rules is well understood and implemented for positive queries, in particular conjunctive queries. For queries with negation, however, there is no agreed-upon semantics or standard implementation. This problem is unknown for simpler rule languages, such as Datalog, where it is intuitive and practical to evaluate negative queries over the least model. This fails for existential rules, which instead of a single least model have multiple universal models that may not lead to the same results for negative queries. We therefore propose universal core models as a basis for a meaningful (non-monotonic) semantics for queries with negation. Since cores are hard to compute, we identify syntactic conditions (on rules and queries) under which our core-based semantics can equivalently be obtained for other universal models, such as those produced by practical chase algorithms. Finally, we use our findings to propose a semantics for a broad class of existential rules with negation.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

We consider first-order (FO) rewritability for query answering in ontology mediated querying (OMQ) in which ontologies are formulated in Horn fragments of description logics (DLs). In general, OMQ approaches for such logics rely on non-FO rewriting of the query and/or on non-FO completion of the data, called a ABox. Specifically, we consider the problem of FO rewritability in terms of Beth definability, and show how Craig interpolation can then be used to effectively construct the rewritings, when they exist, from the Clark’s completion of Datalog-like programs encoding a given DL TBox and optionally a query. We show how this approach to FO rewritability can also be used to (a) capture integrity constraints commonly available in backend relational data sources, (b) capture constraints inherent in mapping such sources to an ABox , and (c) can be used an alternative to deriving so-called perfect rewritings of queries in the case of DL-Lite ontologies.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

We present a new algorithm A*+BFHS for solving problems with unit-cost operators where A* and IDA* fail due to memory limitations and/or the existence of many distinct paths between the same pair of nodes. A*+BFHS is based on A* and breadth-first heuristic search (BFHS). A*+BFHS combines advantages from both algorithms, namely A*'s node ordering, BFHS's memory savings, and both algorithms' duplicate detection. On easy problems, A*+BFHS behaves the same as A*. On hard problems, it is slower than A* but saves a large amount of memory. Compared to BFIDA*, A*+BFHS reduces the search time and/or memory requirement by several times on a variety of planning domains.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

How deep neural networks (DNNs) learn from noisy labels has been studied extensively in image classification but much less in image segmentation. So far, our understanding of the learning behavior of DNNs trained by noisy segmentation labels remains limited. In this study, we address this deficiency in both binary segmentation of biological microscopy images and multi-class segmentation of natural images. We generate extremely noisy labels by randomly sampling a small fraction (e.g., 10%) or flipping a large fraction (e.g., 90%) of the ground truth labels. When trained with these noisy labels, DNNs provide largely the same segmentation performance as trained by the original ground truth. This indicates that DNNs learn structures hidden in labels rather than pixel-level labels per se in their supervised training for semantic segmentation. We refer to these hidden structures in labels as meta-structures. When DNNs are trained by labels with different perturbations to the meta-structure, we find consistent degradation in their segmentation performance. In contrast, incorporation of meta-structure information substantially improves performance of an unsupervised segmentation model developed for binary semantic segmentation. We define meta-structures mathematically as spatial density distributions and show both theoretically and experimentally how this formulation explains key observed learning behavior of DNNs.

poster session 4 @ red 3, poster session 8 @ red 3, oral session 8 @ red 3, poster session 4, poster session 8, oral session 8

The Latin square completion (LSC) problem is an important NP-complete problem with numerous applications. Given its theoretical and practical importance, several algorithms are designed for solving the LSC problem. In this work, to further improve the performance, a fast local search algorithm is developed based on three main ideas. Firstly, a reduction reasoning technique is used to reduce the scale of search space. Secondly, we propose a novel conflict value selection heuristic, which considers the history conflicting information of vertices as a selection criterion when more than one vertex have equal values on the primary scoring function. Thirdly, during the search phase, we record previous history search information and then make use of these information to restart the candidate solution. Experimental results show that our proposed algorithm significantly outperforms the state-of-the-art heuristic algorithms on almost all instances in terms of success rate and run time.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

In humans, perceptual awareness facilitates the fast recognition and extraction of information from sensory input. This awareness largely depends on how the human agent interacts with the environment. In this work, we propose active neural generative coding, a computational framework for learning action-driven generative models without backpropagation of errors (backprop) in dynamic environments. Specifically, we develop an intelligent agent that operates even with sparse rewards, drawing inspiration from the cognitive theory of planning as inference. We demonstrate on several simple control problems that our framework performs competitively with deep Q-learning. The robust performance of our agent offers promising evidence that a backprop-free approach for neural inference and learning can drive goal-directed behavior.

poster session 4 @ red 4, poster session 11 @ red 4, poster session 4, poster session 11

Subject-invariant facial action unit (AU) recognition remains challenging for the reason that the data distribution varies among subjects. In this paper, we propose a causal inference framework for subject-invariant facial action unit recognition. To illustrate the causal effect existing in AU recognition task, we formulate the causalities among facial images, subjects, latent AU semantic relations, and estimated AU occurrence probabilities via a structural causal model. By constructing such a causal diagram, we clarify the causal-effect among variables and propose a plug-in causal intervention module, CIS, to deconfound the confounder Subject in the causal diagram. Extensive experiments conducted on two commonly used AU benchmark datasets, BP4D and DISFA, show the effectiveness of our CIS, and the model with CIS inserted, CISNet, has achieved state-of-the-art performance.

poster session 4 @ red 1, poster session 7 @ red 1, oral session 7 @ red 1, poster session 4, poster session 7, oral session 7

Despite the success of deep neural networks (DNNs) for real-world applications over time-series data such as mobile health, little is known about how to train robust DNNs for time-series domain due to its unique characteristics compared to images and text data. In this paper, we fill this gap by proposing a novel algorithmic framework referred as RObust Training for Time-Series (ROTS) to create robust deep models for time-series classification tasks. We formulate a min-max optimization problem over the model parameters by explicitly reasoning about the robustness criteria in terms of additive perturbations to time-series inputs measured by the global alignment kernel (GAK) based distance. This problem is an instance of the family of compositional min-max optimization problems, which are hard to solve. We propose a principled stochastic compositional alternating gradient descent ascent (SCAGDA) algorithm for this family of optimization problems. Unlike traditional methods for time-series that require approximate computation of distance measures, SCAGDA approximates the GAK based distance on-the-fly using a moving average approach. We theoretically analyze the convergence rate of SCAGDA and provide strong theoretical support for the estimation of GAK based distance. One advantage of our formulation is that we naturally obtain an instantiation of ROTS with dynamic time warping (DTW) based distance. Our experiments on real-world benchmarks demonstrate that ROTS creates more robust deep models when compared to adversarial training using prior methods that rely on data augmentation or new definitions of loss functions. We also demonstrate the importance of GAK and DTW for time-series data over the Euclidean distance.

poster session 4 @ red 4, poster session 11 @ red 4, poster session 4, poster session 11

Centralized Training for Decentralized Execution, where training is done in a centralized offline fashion, has become a popular solution paradigm in Multi-Agent Reinforcement Learning. Many such methods take the form of actor-critic with state-based critics, since centralized training allows access to the true system state, which can be useful during training despite not being available at execution time. State-based critics have become a common empirical choice, albeit one which has had limited theoretical justification or analysis. In this paper, we show that state-based critics can introduce bias in the policy gradient estimates, potentially undermining the asymptotic guarantees of the algorithm. We also show that, even if the state-based critics do not introduce any bias, they can still result in a larger gradient variance, contrary to the common intuition. Finally, we show the effects of the theories in practice by comparing different forms of centralized critics on a wide range of common benchmarks, and detail how various environmental properties are related to the effectiveness of different types of critics.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

Obtaining a well-trained model involves expensive data collection and training procedures, therefore the model is a valuable intellectual property. Recent studies revealed that adversaries can `steal' deployed models even when they have no training samples and can not get access to the model parameters or structures. Currently, there were some defense methods to alleviate this threat, mostly by increasing the cost of model stealing. In this paper, we explore the defense from another angle by verifying whether a suspicious model contains the knowledge of defender-specified \emph{external features}. Specifically, we embed the external features by tempering a few training samples with style transfer. We then train a meta-classifier to determine whether a model is stolen from the victim. This approach is inspired by the understanding that the stolen models should contain the knowledge of features learned by the victim model. We examine our method on both CIFAR-10 and ImageNet datasets. Experimental results demonstrate that our method is effective in detecting different types of model stealing simultaneously, even if the stolen model is obtained via a multi-stage stealing process. The codes for reproducing main results are available at Github (https://github.com/zlh-thu/StealingVerification).

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

Neighborhood gentrification plays a significant role in shaping the social and economic status of both individuals and communities. While some efforts have been made to detect gentrification in cities, existing approaches mainly relies on estimated measures from survey data and requires substantial work of human labeling yet fails to characterize the physical appearance of neighborhoods. To this end, we introduce a novel approach to incorporate data like street view images and POI features to represent urban neighborhoods comprehensively at each timestamp. We show the effectiveness of the proposed methods with previous research on gentrification measures: each neighborhood representation we trained not only indicates its gentrification status, but also could become supplementary parts for the current measures and valid resource for researchers and policy makers.

poster session 4 @ blue 2, poster session 7 @ blue 2, poster session 4, poster session 7

Multi-Agent Path Finding (MAPF) is the problem of planning collision-free paths for multiple agents in a shared environment. In this paper, we propose a novel algorithm MAPF-LNS2 based on large neighborhood search for solving MAPF efficiently. Starting from a set of paths that contains collisions, MAPF-LNS2 repeatedly selects a subset of colliding agents and replans their paths to reduce the number of collisions until the paths become collision-free. We compare MAPF-LNS2 against a variety of state-of-the-art MAPF algorithms, including Prioritized Planning with random restarts, EECBS, and PPS, and show that MAPF-LNS2 runs significantly faster than them while still providing near-optimal solutions in most cases. With a time limit of just 5 minutes, MAPF-LNS2 managed to solve 80% of the random-scenario instances on all 33 maps with the largest numbers of agents from the MAPF benchmark, which clearly defines the new state-of-the-art suboptimal MAPF algorithm.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

This study focuses on the Nesterov's accelerated quasi-Newton (NAQ) method in the context of deep neural networks (DNN) and its applications. The thesis objective is to confirm the robustness and efficiency of Nesterov's acceleration to quasi-Netwon (QN) methods by developing practical algorithms for different fields of optimization problems.

poster session 3 @ red 4, poster session 4 @ red 4, poster session 3, poster session 4

We introduce a Recursive INsertion-based Encoder (RINE), a novel approach for semantic parsing in task-oriented dialog. Our model consists of an encoder network that incrementally builds the semantic parse tree by predicting the non-terminal label and its positions in the linearized tree. At the generation time, the model constructs the semantic parse tree by recursively inserting the predicted non-terminal labels at the predicted positions until termination. RINE achieves state-of-the-art exact match accuracy on low- and high-resource versions of the conversational semantic parsing benchmark TOP (Gupta et al., 2018; Chen et al., 2020), outperforming strong sequence-to-sequence models and transition-based parsers. We also show that our model design is applicable to nested named entity recognition task, where it performs on par with state-of-the-art approach designed for that task. Finally, we demonstrate that our approach is 2-3.5 times faster than the sequence-to-sequence model at inference time.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

Improving health equity is an urgent task for our society. The advent of mobile clinics plays an important role in enhancing health equity, as they can provide easier access to preventive healthcare for patients from disadvantaged populations. For effective functioning of mobile clinics, accurate prediction of demand (expected number of individuals visiting mobile clinic) is the key to their daily operations and staff/resource allocation. Despite its importance, there have been very limited studies on predicting demand of mobile clinics. To the best of our knowledge, we are among the first to explore this area, using AI-based techniques. A crucial challenge in this task is that there are no known existing data sources from which we can extract useful information to account for the exogenous factors that may affect the demand, while considering protection of client privacy. We propose a novel methodology that completely uses public data sources to extract the features, with several new components that are designed to improve the prediction. Empirical evaluation on a real-world dataset from the mobile clinic The Family Van shows that, by leveraging publicly available data (which introduces no extra monetary cost to the mobile clinics), our AI-based method achieves 26.4% - 51.8% lower Root Mean Squared Error (RMSE) than the historical average-based estimation (which is presently employed by mobile clinics like The Family Van). Our algorithm makes it possible for mobile clinics to plan proactively, rather than reactively, as what has been doing.

poster session 4 @ red 6, poster session 9 @ red 6, poster session 4, poster session 9

The maximum $k$-club problem (M$k$CP) is an important clique relaxation problem with wide applications. Previous M$k$CP algorithms only work on small-scale instances and are not applicable for large-scale instances. For solving instances with different scales, this paper develops an efficient local search algorithm named Nu$k$CP for the M$k$CP which mainly includes two novel ideas. First, we propose a dynamic reduction strategy, which makes a good balance between the time efficiency and the precision effectiveness of the upper bound calculation. Second, a stratified threshold configuration checking strategy is designed by giving different priorities for the neighborhood in the different levels. Experiments on a broad range of different scale instances show that Nu$k$CP significantly outperforms the state-of-the-art M$k$CP algorithms on most instances.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

Modeling the preferences of agents over a set of alternatives is a principal concern in many areas. The dominant approach has been to find a single reward/utility function with the property that alternatives yielding higher rewards are preferred over alternatives yielding lower rewards. However, in many settings, preferences are based on multiple—often competing—objectives; a single reward function is not adequate to represent such preferences. This paper proposes a method for inferring multi-objective reward-based representations of an agent's observed preferences. We model the agent's priorities over different objectives as entering lexicographically, so that objectives with lower priorities matter only when the agent is indifferent with respect to objectives with higher priorities. We offer two example applications in healthcare—one inspired by cancer treatment, the other inspired by organ transplantation—to illustrate how the lexicographically-ordered rewards we learn can provide a better understanding of a decision-maker's preferences and help improve policies when used in reinforcement learning.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

In bounded-suboptimal heuristic search, one attempts to find a solution that costs no more than a prespecified factor of optimal as quickly as possible. This is an important setting, as it admits faster-than-optimal solving while retaining some control over solution cost. In this paper, we investigate several new algorithms for bounded-suboptimal search, including novel variants of EES and DPS, the two most prominent previous proposals, and methods inspired by recent work in bounded-cost search that leverages uncertainty estimates of the heuristic. We perform what is, to our knowledge, the most comprehensive empirical comparison of bounded-suboptimal search algorithms to date, including both search and planning benchmarks, and we find that one of the new algorithms, a simple alternating queue scheme, significantly outperforms previous work.

poster session 4 @ blue 4, poster session 8 @ blue 4, oral session 4 @ blue 4, poster session 4, poster session 8, oral session 4

Unsupervised image-to-image (UI2I) translation methods aim to learn a mapping between different visual domains with well-preserved content and consistent structure. It has been proven that the generated images are quite useful for enhancing the performance of computer vision tasks like object detection in a different domain with distribution discrepancies. Current methods require large amounts of images in both source and target domains for successful translation. However, data collection and annotations in many scenarios are infeasible or even impossible. In this paper, we propose an Object-Aware Few-Shot UI2I Translation (OA-FSUI2IT) framework to address the few-shot cross domain (FSCD) object detection task with limited unlabeled images in the target domain. To this end, we first introduce a discriminator augmentation (DA) module into the OA-FSUI2IT framework for successful few-shot UI2I translation. Then, we present a patch pyramid contrastive learning (PPCL) strategy to further improve the quality of the generated images. Last, we propose a self-supervised content-consistency (SSCC) loss to enforce the content-consistency in the translation. We implement extensive experiments to demonstrate the effectiveness of our OA-FSUI2IT framework for FSCD object detection and achieve state-of-the-art performance on the benchmarks of Normal-to-Foggy, Day-to-Night, and Cross-scene adaptation. The source code of our proposed method is also available at https://github.com/emdata-ailab/FSCD-Det.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

The Moran process is a classic stochastic process that models invasion dynamics on graphs. A single ``mutant'' (e.g., a new opinion, strain, social trait etc.) invades a population of residents spread over the nodes of a graph. The mutant fitness advantage $\delta\geq 0$ determines how aggressively mutants propagate to their neighbors. The quantity of interest is the fixation probability, i.e., the probability that the initial mutant eventually takes over the whole population. However, in realistic settings, the invading mutant has an advantage only in certain locations. E.g., the ability to metabolize a certain sugar is an advantageous trait to bacteria only when the sugar is actually present in their surroundings. In this paper we introduce the positional Moran process, a natural generalization in which the mutant fitness advantage is only realized on specific nodes called active nodes, and study the problem of fixation maximization: given a budget $k$, choose a set of $k$ active nodes that maximize the fixation probability of the invading mutant. We show that the problem is NP-hard, while the optimization function is not submodular, thus indicating strong computational hardness. We focus on two natural limits. In the limit of $\delta\to\infty$ (strong selection), although the problem remains NP-hard, the optimization function becomes submodular and thus admits a constant-factor approximation using a simple greedy algorithm. In the limit of $\delta\to 0$ (weak selection), we show that we can obtain a tight approximation in $O(n^{2\cdot \omega})$ time, where $\omega$ is the matrix-multiplication exponent. An experimental evaluation of the new algorithms along with some proposed heuristics corroborates our results.

poster session 4 @ blue 6, poster session 9 @ blue 6, oral session 9 @ blue 6, poster session 4, poster session 9, oral session 9

Configuration checking (CC) has been confirmed to alleviate the cycling problem in local search for combinatorial optimization problems (COPs). When using CC heuristics in local search for graph problems, a critical concept is the configuration of the vertices. All existing CC variants employ either 1- or 2-level neighborhoods of a vertex as its configuration. Inspired by the idea that neighborhoods with different levels should have different contributions to solving COPs, we propose the probabilistic configuration (PC), which introduces probabilities for neighborhoods at different levels to consider the impact of neighborhoods of different levels on the CC strategy. Based on the concept of PC, we first propose probabilistic configuration checking (PCC), which can be developed in an automated and lightweight favor. We then apply PCC to two classic COPs which have been shown to achieve good results by using CC, and our preliminary results confirm that PCC improves the existing algorithms because PCC alleviates the cycling problem.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

Understanding users' behavior via IP addresses is essential towards numerous practical IP-based applications such as online content delivery, fraud prevention, and many others. Among which profiling IP address has been extensively studied, such as IP geolocation and anomaly detection. However, less is known about the scenario of an IP address, e.g., dedicated enterprise network or home broadband. In this work, we initiate the first attempt to address a large-scale IP scenario prediction problem. Specifically, we collect IP scenario data from four regions and propose a novel deep ensemble learning-based model to learn IP assignment rules and complex feature interactions. Extensive experiments support that our method can make accurate IP scenario identification and generalize from data in one region to another.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

Convolutional neural networks based single-image super-resolution (SISR) has made great progress in recent years. However, it is difficult to apply these methods to real-world scenarios due to the computational and memory cost. Meanwhile, how to take full advantage of the intermediate features under the constraints of limited parameters and calculations is also a huge challenge. To alleviate these issues, we propose a lightweight and efficient Feature Distillation Interaction Weighted Network (FDIWN). Specifically, FDIWN utilizes a series of specially designed Feature Shuffle Weighted Groups (FSWG) as the backbone, and several novel mutual Wide-residual Distillation Interaction Blocks (WDIB) form an FSWG. In addition, Wide Identical Residual Weighting (WIRW) units and Wide Convolutional Residual Weighting (WCRW) units are introduced into WDIB for better feature distillation. Moreover, a Wide-Residual Distillation Connection (WRDC) framework and a Self-Calibration Fusion (SCF) unit are proposed to interact features with different scales more flexibly and efficiently. Extensive experiments show that our FDIWN is superior to other models to strike a good balance between model performance and efficiency. The code is available at https://github.com/IVIPLab/FDIWN.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

Anomaly detection attempts to find examples in a dataset that do not conform to the expected behaviour. Algorithms for this task assign an anomaly score to each example representing its degree of anomalousness. Setting a threshold on the anomaly scores enables converting these scores into a discrete prediction for each example. Setting an appropriate threshold is challenging in practice since anomaly detection is often treated as an unsupervised problem. A common approach is to set the threshold based on the dataset's contamination factor, i.e., the proportion of anomalous examples in the data. While the contamination factor may be known based on domain knowledge, it is often necessary to estimate it by labeling data. However, many anomaly detection problems involve monitoring multiple related, yet slightly different entities (e.g., a fleet of machines). Then, estimating the contamination factor for each dataset separately by labeling data would be extremely time-consuming. Therefore, this paper introduces a method for transferring the known contamination factor from one dataset (the source domain) to a related dataset where it is unknown (the target domain). Our approach does not require labeled target data and is based on modelling the shape of the distribution of the anomaly scores in both domains. We theoretically analyze how our method behaves when the (biased) target domain anomaly score distribution converges to its true one. Empirically, we demonstrate the effectiveness of our method on real-world datasets showing that it outperforms several baselines.

poster session 4 @ blue 2, poster session 7 @ blue 2, poster session 4, poster session 7

Neural networks have a bias towards low frequency functions. This spectral bias has been the subject of several previous studies, both empirical and theoretical. Here we present a computable definition of the spectral bias based on a decomposition of the reconstruction error into a low and a high frequency component. The distinction between low and high frequencies is made in a way that allows for easy interpretation of the spectral bias. Furthermore, we present two methods for estimating the spectral bias. Method 1 relies on the use of the discrete Fourier transform to explicitly estimate the Fourier spectrum of the prediction residual, and Method 2 uses convolution to extract the low frequency components, where the convolution integral is estimated by Monte Carlo methods. The spectral bias depends on the distribution of the data, which is approximated with kernel density estimation when unknown. We devise a set of numerical experiments that confirm that low frequencies are learned first, a behavior quantified by our definition.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

Malicious applications of deepfakes (i.e., technologies generating target facial attributes or entire faces from facial images) have posed a huge threat to individuals' reputation and security. To mitigate these threats, recent studies have proposed adversarial watermarks to combat deepfake models, leading them to generate distorted outputs. Despite achieving impressive results, these adversarial watermarks have low image-level and model-level transferability, meaning that they can protect only one facial image from one specific deepfake model. To address these issues, we propose a novel solution that can generate a Cross-Model Universal Adversarial Watermark (CMUA-Watermark), protecting a large number of facial images from multiple deepfake models. Specifically, we begin by proposing a cross-model universal attack pipeline that attacks multiple deepfake models iteratively. Then, we design a two-level perturbation fusion strategy to alleviate the conflict between the adversarial watermarks generated by different facial images and models. Moreover, we address the key problem in cross-model optimization with a heuristic approach to automatically find the suitable attack step sizes for different models, further weakening the model-level conflict. Finally, we introduce a more reasonable and comprehensive evaluation method to fully test the proposed method and compare it with existing ones. Extensive experimental results demonstrate that the proposed CMUA-Watermark can effectively distort the fake facial images generated by multiple deepfake models while achieving a better performance than existing methods.

poster session 4 @ red 2, poster session 11 @ red 2, poster session 4, poster session 11

In recent years, creative content generations like style transfer and neural photo editing have attracted more and more attention. Among these, cartoonization of real-world scenes has promising applications in entertainment and industry. Different from image translations focusing on improving the style effect of generated images, video cartoonization has additional requirements on the temporal consistency. In this paper, we propose a spatially-adaptive semantic alignment framework with perceptual motion consistency for coherent video cartoonization in an unsupervised manner. The semantic alignment module is designed to restore deformation of semantic structure caused by spatial information lost in the encoder-decoder architecture. Furthermore, we introduce the spatio-temporal correlative map as a style-independent, global-aware regularization on perceptual motion consistency. Deriving from similarity measurement of high-level features in photo and cartoon frames, it captures global semantic information beyond raw pixel-value of optical flow. Besides, the similarity measurement disentangles temporal relationship from domain-specific style properties, which helps regularize the temporal consistency without hurting style effects of cartoon images. Qualitative and quantitative experiments demonstrate our method is able to generate highly stylistic and temporal consistent cartoon videos.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

We introduce a new statistical verification algorithm that formally quantifies the behavioral robustness of any time-continuous process formulated as a continuous-depth model. Our algorithm solves a set of global optimization (Go) problems over a given time horizon to construct a tight enclosure (Tube) of the set of all process executions starting from a ball of initial states. We call our algorithm GoTube. Through its construction, GoTube ensures that the bounding tube is conservative up to a desired probability and up to a desired tightness. GoTube is implemented in JAX and optimized to scale to complex continuous-depth neural network models. Compared to advanced reachability analysis tools for time-continuous neural networks, GoTube does not accumulate overapproximation errors between time steps and avoids the infamous wrapping effect inherent in symbolic techniques. We show that GoTube substantially outperforms state-of-the-art verification tools in terms of the size of the initial ball, speed, time-horizon, task completion, and scalability on a large set of experiments. GoTube is stable and sets the state-of-the-art in terms of its ability to scale to time horizons well beyond what has been previously possible.

poster session 4 @ blue 5, poster session 8 @ blue 5, oral session 4 @ blue 5, poster session 4, poster session 8, oral session 4

Transformer models have achieved promising results on natural language processing (NLP) tasks including extractive question answering (QA). Common Transformer encoders used in NLP tasks process the hidden states of all input tokens in the context paragraph throughout all layers. However, different from other tasks such as sequence classification, answering the raised question does not necessarily need all the tokens in the context paragraph. Following this motivation, we propose Block-skim, which learns to skim unnecessary context in higher hidden layers to improve and accelerate the Transformer performance. The key idea of Block-Skim is to identify the context that must be further processed and those that could be safely discarded early on during inference. Critically, we find that such information could be sufficiently derived from the self-attention weights inside the Transformer model. We further prune the hidden states corresponding to the unnecessary positions early in lower layers, achieving significant inference-time speedup. To our surprise, we observe that models pruned in this way outperform their full-size counterparts. Block-Skim improves QA models' accuracy on different datasets and achieves 3 times speedup on BERT-base model.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

Micronutrient deficiency (MND), which is a form of malnutrition that can have serious health consequences, is difficult to diagnose in early stages without blood draws, which are expensive and time-consuming to collect and process. It is even more difficult at a public health scale seeking to identify regions at higher risk of MND. To provide data more widely and frequently, we propose an accurate, scalable, low-cost, and interpretable regional-level MND prediction system. Specifically, our work is the first to use satellite data, such as forest cover, weather, and presence of water, to predict deficiency of micronutrients such as iron, Vitamin B12, and Vitamin A, directly from their biomarkers. We use real-world, ground truth biomarker data collected from four different regions across Madagascar for training, and demonstrate that satellite data are viable for predicting regional-level MND, surprisingly exceeding the performance of baseline predictions based only on survey responses. Our method could be broadly applied to other countries where satellite data are available, and potentially create high societal impact if these predictions are used by policy makers, public health officials, or healthcare providers.

poster session 4 @ red 6, poster session 9 @ red 6, poster session 4, poster session 9

The Linear Threshold Model is a widely used model that describes how information diffuses through a social network. According to this model, an individual adopts an idea or product after the proportion of their neighbors who have adopted it reaches a certain threshold. Typical applications of the Linear Threshold Model assume that thresholds are either the same for all network nodes or randomly distributed, even though some people may be more susceptible to peer pressure than others. To address individual-level differences, we propose causal inference methods for estimating individual thresholds that can more accurately predict whether and when individuals will be affected by their peers. We introduce the concept of heterogeneous peer effects and develop a Structural Causal Model which corresponds to the Linear Threshold Model and supports heterogeneous peer effect identification and estimation. We develop two algorithms for individual threshold estimation, one based on causal trees and one based on causal meta-learners. Our experimental results on synthetic and real- world datasets show that our proposed models can better predict individual-level thresholds in the Linear Threshold Model and thus more precisely predict which nodes will get activated over time.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

Skeleton-based action recognition models have recently been shown to be vulnerable to adversarial attacks. Compared to adversarial attacks on images, perturbations to skeletons are typically bounded to a lower dimension of approximately 100 per frame. This lower-dimensional setting makes it more difficult to generate imperceptible perturbations. Existing attacks resolve this by exploiting the temporal structure of the skeleton motion so that the perturbation dimension increases to thousands. In this paper, we show that adversarial attacks can be performed on skeleton-based action recognition models, even in a significantly low-dimensional setting without any temporal manipulation. Specifically, we restrict the perturbations to the lengths of the skeleton's bones, which allows an adversary to manipulate only approximately 30 effective dimensions. We conducted experiments on the NTU RGB+D and HDM05 datasets and demonstrate that the proposed attack successfully deceived models with sometimes greater than 90\% success rate by small perturbations. Furthermore, we discovered an interesting phenomenon: in our low-dimensional setting, the adversarial training with the bone length attack shares a similar property with data augmentation, and it not only improves the adversarial robustness but also improves the classification accuracy on the original data. This is an interesting counterexample of the trade-off between adversarial robustness and clean accuracy, which has been widely observed in studies on adversarial training in the high-dimensional regime.

poster session 4 @ blue 6, poster session 9 @ blue 6, poster session 4, poster session 9

The paper addresses the challenge of accelerating identification of changes in risk drivers in the insurance industry. Specifically, the work presents a method to identify significant news events ("signals") from batches of news data to inform Life & Health insurance decisions. Signals are defined as events that are relevant to a tracked risk driver, widely discussed in multiple news outlets, contain novel information and affect stakeholders. The method converts unstructured data (news articles) into a sequence of keywords by employing a linguistic knowledge graph-based model. Then, for each time window, the method forms a graph with extracted keywords as nodes and draws weighted edges based on keyword co-occurrences in articles. Lastly, events are derived in an unsupervised way as graph communities and scored for the requirements of a signal: relevance, novelty and virality. The methodology is illustrated for a Life & Health topic using news articles from Dow Jones DNA proprietary data set, and assessed against baselines on a publicly available news data set. The method is implemented as an analytics engine in Early Warning System deployed at Swiss Re for the last 1.5 years to extract relevant events from live news data. We present the system's architectural design in production and discuss its use and impact.

poster session 4 @ red 6, poster session 8 @ red 6, poster session 4, poster session 8

Expert decision makers are starting to rely on data-driven automated agents to assist them with various tasks. For this collaboration to perform properly, the human decision maker must have a mental model of when and when not to rely on the agent. In this work, we aim to ensure that human decision makers learn a valid mental model of the agent's strengths and weaknesses. To accomplish this goal, we propose an exemplar-based teaching strategy where humans solve a set of selected examples and with our help generalize from them to the domain. We present a novel parameterization of the human's mental model of the AI that applies a nearest neighbor rule in local regions surrounding the teaching examples. Using this model, we derive a near-optimal strategy for selecting a representative teaching set. We validate the benefits of our teaching strategy on a multi-hop question answering task with an interpretable AI model using crowd workers. We find that when workers draw the right lessons from the teaching stage, their task performance improves. We furthermore validate our method on a set of synthetic experiments.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

Common sense has always been of interest in AI, but has rarely taken center stage. Despite its mention in one of John McCarthy's earliest papers and years of work by dedicated researchers, arguably no AI system with a serious amount of general common sense has ever emerged. Why is that? What's missing? Examples of AI systems' failures of common sense abound, and they point to AI's frequent focus on expertise as the cause. Those attempting to break the brittleness barrier, even in the context of modern deep learning, have tended to invest their energy in large numbers of small bits of commonsense knowledge. But all the commonsense knowledge fragments in the world don't add up to a system that actually demonstrates common sense in a human-like way. We advocate examining common sense from a broader perspective than in the past. Common sense is more complex than it has been taken to be and is worthy of its own scientific exploration.

poster session 4 @ blue 1, poster session 11 @ blue 1, oral session 11 @ blue 1, poster session 4, poster session 11, oral session 11

Generalized planning accelerates classical planning by finding an algorithm-like policy that solves multiple instances of a task. A generalized plan can be learned from a few training examples and applied to an entire domain of problems. Generalized planning approaches perform well in discrete AI planning problems that involve large numbers of objects and extended action sequences to achieve the goal. In this paper, we propose an algorithm for learning features, abstractions, and generalized plans for continuous robotic task and motion planning (TAMP) and examine the unique difficulties that arise when forced to consider geometric and physical constraints as a part of the generalized plan. Additionally, we show that these simple generalized plans learned from only a handful of examples can be used to improve the search efficiency of TAMP solvers.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

Methods for Knowledge Base Completion (KBC) reason about a knowledge base (KB) in order to derive new facts that should be included in the KB. This is challenging for two reasons. First, KBs only contain positive examples. This complicates model evaluation which needs both positive and negative examples. Second, those facts that were selected to be included in the knowledge base, are most likely not an i.i.d. sample of the true facts, due to the way knowledge bases are constructed. In this paper, we focus on rule-based approaches, which traditionally address the first challenge by making assumptions that enable identifying negative examples, which in turn makes it possible to compute a rule's confidence or precision. However, they largely ignore the second challenge, which means that their estimates of a rule's confidence can be biased. This paper approaches rule-based KBC through the lens of PU-learning, which can cope with both challenges. We make three contributions.: (1) We provide a unifying view that formalizes the relationship between multiple existing confidences measures based on (i) what assumption they make about and (ii) how their accuracy depends on the selection mechanism. (2) We introduce two new confidence measures that can mitigate known biases by using propensity scores that quantify how likely a fact is to be included the KB. (3) We show through theoretical and empirical analysis that taking the bias into account improves the confidence estimates, even when the propensity scores are not known exactly.

poster session 4 @ blue 2, poster session 7 @ blue 2, oral session 4 @ blue 2, poster session 4, poster session 7, oral session 4

Networked discrete dynamical systems are commonly used as formal models to study the spread of contagions over networks. Fixed points of such dynamical systems represent configurations to which systems may converge. In the case of undesirable contagions (such as epidemics, fake news and rumors) spreading over social networks (where nodes represent a population), convergence to configurations with a small number of affected nodes is highly desirable. Motivated by such considerations, we formulate a novel optimization problem for discrete dynamical systems, where the state of each node is from {0, 1} and state 1 indicates a node that has been infected by the contagion. Specifically, the goal of the problem is to find a fixed point of such a system where the number of nodes in state 1 is a minimum. We also require such a fixed point to be nontrivial (to avoid the trivial case where all nodes are in state 0). Our results are for dynamical systems where the behavior of each node is specified by a threshold function. These functions capture simple as well as complex contagions and have also been studied in game theory. We show that, unless P = NP, there is no polynomial time algorithm for approximating the solution to within the factor n^{1 - epsilon} for any constant \epsilon > 0, even when the underlying network is bipartite. We then explore several approaches to cope with the computational intractability of the problem. First, we identify several special cases (e.g., dynamical systems over directed acyclic graphs, progressive contagions where nodes cannot transition from state 1 to state 0) for which the problem can be solved efficiently. We also identify a version that is fixed parameter tractable with respect to a structural parameter associated with the problem. Next, we develop an integer linear programming (ILP) formulation for the problem so that existing solvers such as Gurobi can be used to solve the problem in practice for networks of reasonable size. For larger networks, we propose a general heuristic framework for the problem. We present extensive experimental results using real-world networks to demonstrate the effectiveness of the proposed heuristic framework.

poster session 4 @ red 3, poster session 8 @ red 3, oral session 4 @ red 3, poster session 4, poster session 8, oral session 4

As the labeling cost for different modules in task-oriented dialog (ToD) systems is high, a major challenge is to learn different tasks with the least amount of labeled data. Recently, pre-trained language models (PLMs) have shown promising results for few-shot learning in ToD. To better utilize the power of PLMs, this paper proposes Comprehensive Instruction (CINS) that exploits PLMs with extra task-specific instructions. We design a schema (definition, constraint, prompt) of instructions and their customized realizations for three important downstream tasks in ToD, ie. intent classification, dialog state tracking, and natural language generation. A sequence-to-sequence model (T5) is adopted to solve these three tasks in a unified framework. Extensive experiments are conducted on these ToD tasks in realistic few-shot learning scenarios with small validation data. Empirical results demonstrate that the proposed CINS approach consistently improves techniques that finetune PLMs with raw input or short prompt.

poster session 4 @ red 5, poster session 11 @ red 5, oral session 11 @ red 5, poster session 4, poster session 11, oral session 11

In this paper, we propose on-device voice command assistants for mobile games to increase user experiences even in hands-busy situations such as driving and cooking. Since most of the current mobile games cost large memory (e.g. more than 1GB memory), so it is necessary to reduce memory usage further to integrate voice commands systems on mobile clients. Therefore a need to design an on-device automatic speech recognition system that costs minimal memory and CPU resources rises. To this end, we apply cross layer parameter sharing to Conformer, named MONICA2 which results in lower memory usage for on-device speech recognition. MONICA2 reduces the number of parameters of deep neural network by 58%, with minimal recognition accuracy degradation measured in word error rate on Librispeech benchmark. As an on-device voice command user interface, MONICA2 costs only 12.8MB mobile memory and the average inference time for 3-seconds voice command is about 30ms, which is profiled in Samsung Galaxy S9. As far as we know, MONICA2 is the most memory efficient yet accurate on-device speech recognition which could be applied to various applications such as mobile games, IoT devices, etc.

poster session 4 @ blue 1, poster session 11 @ blue 1, poster session 4, poster session 11

We address efficient calculation of influence functions (Koh & Liang, 2017) for tracking predictions back to the training data. We propose and analyze a new approach to speeding up the inverse Hessian calculation based on Arnoldi iteration (Arnoldi, 1951). With this improvement, we achieve, to the best of our knowledge, the first successful implementation of influence functions that scales to full-size (language and vision) Transformer models with several hundreds of millions of parameters. We evaluate our approach in image classification and sequence-to-sequence tasks with tens to a hundred of millions of training examples. Our code will be available at https://github.com/google-research/jax-influence.

poster session 4 @ blue 2, poster session 7 @ blue 2, poster session 4, poster session 7

Cross-lingual Machine Reading Comprehension (xMRC) is a challenging task due to the lack of training data in low-resource languages. Recent approaches use training data only in a resource-rich language (such as English) to fine-tune large-scale cross-lingual pre-trained language models, which transfer knowledge from resource-rich languages (source) to low-resource languages (target). Due to the big difference between languages, the model fine-tuned only by the source language may not perform well for target languages. In our study, we make an interesting observation that while the top 1 result predicted by the previous approaches may often fail to hit the ground-truth answer, there are still good chances for the correct answer to be contained in the set of top k predicted results. Intuitively, the previous approaches have empowered the model certain level of capability to roughly distinguish good answers from bad ones. However, without sufficient training data, it is not powerful enough to capture the nuances between the accurate answer and those approximate ones. Based on this observation, we develop a two-stage approach to enhance the model performance. The first stage targets at recall; we design a hard-learning (HL) algorithm to maximize the likelihood that the top k predictions contain the accurate answer. The second stage focuses on precision, where an answer-aware contrastive learning (AA-CL) mechanism is developed to learn the minute difference between the accurate answer and other candidates. Extensive experiments show that our model significantly outperforms strong baselines on two cross-lingual MRC benchmark datasets.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

Despite the surprising power of many modern AI systems that often learn their own representations, there is significant discontent about their inscrutability and the attendant problems in their ability to interact with humans. While alternatives such as neuro-symbolic approaches have been proposed, there is a lack of consensus on what they are about. There are often two independent motivations (i) symbols as a lingua franca for human-AI interaction and (ii) symbols as (system-produced) abstractions use in its internal reasoning. The jury is still out on whether AI systems will need to use symbols in their internal reasoning to achieve general intelligence capabilities. Whatever the answer there is, the need for (human-understandable) symbols in human-AI interaction seems quite compelling. Symbols, like emotions, may well not be sine qua non for intelligence per se, but they will be crucial for AI systems to interact with us humans--as we can neither turn off our emotions not get by without our symbols. In particular, in many human-designed domains, humans would be interested in providing explicit (symbolic) knowledge and advice--and expect machine explanations in kind. This alone requires AI systems to at least do their I/O in symbolic terms. In this blue sky paper, we argue this point of view, and discuss research directions that need to be pursued to allow for this type of human-AI interaction.

poster session 4 @ blue 1, poster session 11 @ blue 1, oral session 4 @ blue 1, poster session 4, poster session 11, oral session 4

Deep learning models have shown to be susceptible to universal adversarial perturbation (UAP), which has aroused wide concerns in the community. Compared with the conventional adversarial attacks that generate adversarial samples at the instance level, UAP can fool the target model for different instances with only a single perturbation, enabling us to evaluate the robustness of the model from a more effective and accurate perspective. The existing universal attack methods fail to exploit the differences and connections between the instance and universal levels to produce dominant perturbations. To address this challenge, we propose a new universal attack method that unifies instance-specific and universal attacks from a feature perspective to generate a more dominant UAP. Specifically, we reformulate the UAP generation task as a minimax optimization problem and then utilize the instance-specific attack method to solve the minimization problem thereby obtaining better training data for generating UAP. At the same time, we also introduce a consistency regularizer to explore the relationship between training data, thus further improving the dominance of the generated UAP. Furthermore, our method is generic with no additional assumptions about the training data and hence can be applied to both data-dependent supervised and data-independent unsupervised manners. Extensive experiments demonstrate that the proposed method improves the performance by a significant margin over the existing methods in both data-dependent and data-independent settings.

poster session 4 @ red 2, poster session 11 @ red 2, poster session 4, poster session 11

Demand for image editing has been increasing as users' desire for expression is also increasing. However, for most users, image editing tools are not easy to use since the tools require certain expertise in photo effects and have complex interfaces. Hence, users might need someone to help edit their images, but having a personal dedicated human assistant for every user is impossible to scale. For that reason, an automated assistant system for image editing is desirable. Additionally, users want more image sources for diverse image editing works, and integrating an image search functionality into the editing tool is a potential remedy for this demand. Thus, we propose a dataset of an automated Conversational Agent for Image Search and Editing (CAISE). To our knowledge, this is the first dataset that provides conversational image search and editing annotations, where the agent holds a grounded conversation with users and helps them to search and edit images according to their requests. To build such a system, we first collect image search and editing conversations between pairs of annotators. The assistant-annotators are equipped with a customized image search and editing tool to address the requests from the user-annotators. The functions that the assistant-annotators conduct with the tool are recorded as executable commands, allowing the trained system to be useful for real-world application execution. We also introduce a generator-extractor baseline model for this task, which can adaptively select the source of the next token (i.e., from the vocabulary or from textual/visual contexts) for the executable command. This serves as a strong starting point while still leaving a large human-machine performance gap for useful future work.conversational image search and editing annotations, where the agent holds a grounded conversation with users and helps them to search and edit images according to their requests. To build such a system, we first collect image search and editing conversations between pairs of annotators. The assistant-annotators are equipped with a customized image search and editing tool to address the requests from the user-annotators. The functions that the assistant-annotators conduct with the tool are recorded as executable commands, allowing the trained system to be useful for real-world application execution. We also introduce a generator-extractor baseline model for this task, which can adaptively select the source of the next token (i.e., from the vocabulary or from textual/visual contexts) for the executable command. This serves as a strong starting point while still leaving a large human-machine performance gap for useful future work.

poster session 4 @ red 2, poster session 11 @ red 2, poster session 4, poster session 11

Automated persuasion systems (APS) aim to persuade a user to believe something by entering into a dialogue in which arguments and counterarguments are exchanged. To maximize the probability that an APS is successful in persuading a user, it can identify a global policy that will allow it to select the best arguments it presents at each stage of the dialogue whatever arguments the user presents. However, in real applications, such as for healthcare, it is unlikely the utility of the outcome of the dialogue will be the same, or the exact opposite, for the APS and user. In order to deal with this situation, games in extended form have been harnessed for argumentation in Bi-party Decision Theory. This opens new problems that we address in this paper: (1) How can we use Machine Learning (ML) methods to predict utility functions for different subpopulations of users? and (2) How can we identify for a new user the best utility function from amongst those that we have learned. To this extent, we develop two ML methods, EAI and EDS, that leverage information coming from the users to predict their utilities. EAI is restricted to a fixed amount of information, whereas EDS can choose the information that best detects the subpopulations of a user. We evaluate EAI and EDS in a simulation setting and in a realistic case study concerning healthy eating habits. Results are promising in both cases, but EDS is more effective at predicting useful utility functions.

poster session 4 @ blue 5, poster session 8 @ blue 5, oral session 4 @ blue 5, poster session 4, poster session 8, oral session 4

Answer set programming (ASP) is a popular declarative programming paradigm with a wide range of applications in artificial intelligence. Oftentimes, when modeling an AI problem with ASP, and in particular when we are interested beyond simple search for optimal solutions, an actual solution, differences between solutions, or number of solutions of the ASP program matter. For example, when a user aims to identify a specific answer set according to her needs, or requires the total number of diverging solutions to comprehend probabilistic applications such as reasoning in medical domains. Then, there are only certain problem specific and handcrafted encoding techniques available to navigate the solution space of ASP programs, which is oftentimes not enough. In this paper, we propose a formal and general framework for interactive navigation toward desired subsets of answer sets analogous to faceted browsing. Our approach enables the user to explore the solution space by consciously zooming in or out of sub-spaces of solutions at a certain configurable pace. We illustrate that weighted faceted navigation is computationally hard. Finally, we provide an implementation of our approach that demonstrates the feasibility of our framework for incomprehensible solution spaces.

poster session 4 @ blue 5, poster session 8 @ blue 5, oral session 4 @ blue 5, poster session 4, poster session 8, oral session 4

Text-based games (TBGs) have become a popular proving ground for the demonstration of learning-based agents that make decisions in quasi real-world settings. The crux of the problem for a reinforcement learning agent in such TBGs is identifying the objects in the world, and those objects' relations with that world. While the recent use of text-based resources for increasing an agent's knowledge and improving its generalization have shown promise, we posit in this paper that there is much yet to be learned from visual representations of these same worlds. Specifically, we propose to retrieve images that represent specific instances of text observations from the world and train our agents on such images. This improves the agent's overall understanding of the game scene and objects' relationships to the world around them, and the variety of visual representations on offer allow the agent to generate a better generalization of a relationship. We show that incorporating such images improves the performance of agents in various TBG settings.

poster session 4 @ red 1, poster session 7 @ red 1, oral session 4 @ red 1, poster session 4, poster session 7, oral session 4

Spatio-temporal scene-graph approaches to video-based reasoning tasks such as video question-answering (QA) typically construct such graphs for every video frame. Such approaches often ignore the fact that videos are essentially sequences of 2D ``views'' of events happening in a 3D space, and that the semantics of the 3D scene can thus be carried over from frame to frame. Leveraging this insight, we propose a (2.5+1)D scene graph representation to better capture the spatio-temporal information flows inside the videos. Specifically, we first create a 2.5D (pseudo-3D) scene graph by transforming every 2D frame to have an inferred 3D structure using an off-the-shelf 2D-to-3D transformation module, following which we register the video frames into a shared (2.5+1)D spatio-temporal space and ground each 2D scene graph within it. Such a (2.5+1)D graph is then segregated into a static sub-graph and a dynamic sub-graph, corresponding to whether the objects within them usually move in the world. The nodes in the dynamic graph are enriched with motion features capturing their interactions with other graph nodes. Next, for the video QA task, we present a novel transformer-based reasoning pipeline that embeds the (2.5+1)D graph into a spatio-temporal hierarchical latent space, where the sub-graphs and their interactions are captured at varied granularity. To demonstrate the effectiveness of our approach, we present experiments on the NExT-QA and AVSD-QA datasets. Our results show that our proposed (2.5+1)D representation leads to faster training and inference, while our hierarchical model showcases superior performance on the video QA task versus the state of the art.

poster session 4 @ red 5, poster session 11 @ red 5, oral session 4 @ red 5, poster session 4, poster session 11, oral session 4

Although randomized smoothing has demonstrated high certified robustness and superior scalability to other certified defenses, the high computational overhead of the robustness certification bottlenecks the practical applicability, as it depends heavily on the large sample approximation for estimating the confidence interval. In existing works, the sample size for the confidence interval is universally set and agnostic to the input for prediction. This Input-Agnostic Sampling (IAS) scheme may yield a poor Average Certified Radius (ACR)-runtime trade-off which calls for improvement. In this paper, we propose Input-Specific Sampling (ISS) acceleration to achieve the cost-effectiveness for robustness certification, in an adaptive way of reducing the sampling size based on the input characteristic. Furthermore, our method universally controls the certified radius decline from the ISS sample size reduction. The empirical results on CIFAR-10 and ImageNet show that ISS can speed up the certification by more than three times at a limited cost of 0.05 certified radius. Meanwhile, ISS surpasses IAS on the average certified radius across the extensive hyperparameter settings. Specifically, ISS achieves ACR=0.958 on ImageNet ($\sigma=1.0$) in 250 minutes, compared to ACR=0.917 by IAS under the same condition.

poster session 4 @ red 2, poster session 11 @ red 2, poster session 4, poster session 11

Ontology alignment (a.k.a ontology matching (OM)) plays a critical role in knowledge integration. Owing to the success of machine learning in many domains, it has been applied in OM. However, the existing methods, which often adopt ad-hoc feature engineering or non-contextual word embeddings, have not yet outperformed rule-based systems especially in an unsupervised setting. In this paper, we propose a novel OM system named BERTMap which can support both unsupervised and semi-supervised settings. It first predicts mappings using a classifier based on fine-tuning the contextual embedding model BERT on text semantics corpora extracted from ontologies, and then refines the mappings through extension and repair by utilizing the ontology structure and logic. Our evaluation with three alignment tasks on biomedical ontologies demonstrates that BERTMap can often perform better than the leading OM systems LogMap and AML.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

Off-policy learning plays a pivotal role in optimizing and evaluating policies prior to the online deployment. However, during the real-time serving, we observe varieties of interventions and constraints that cause inconsistency between the online and offline setting, which we summarize and term as runtime uncertainty. Such uncertainty cannot be learned from the logged data due to its abnormality and rareness nature. To assert a certain level of robustness, we perturb the off-policy estimators along an adversarial direction in view of the runtime uncertainty. It allows the resulting estimators to be robust not only to observed but also unexpected runtime uncertainties. Leveraging this idea, we bring runtime-uncertainty robustness to three major off-policy learning methods: the inverse propensity score method, reward-model method, and doubly robust method. We theoretically justify the robustness of our methods to runtime uncertainty, and demonstrate their effectiveness using both the simulation and the real-world online experiments.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

We present a novel parallel algorithm for drawing balanced samples from large populations. When auxiliary variables about the population units are known, balanced sampling improves the quality of the estimations obtained from the sample. Available algorithms, e.g., the cube method, are inherently sequential, and do not scale to large populations. Our parallel algorithm is based on a variant of the cube method for stratified populations. It has the same sample quality as sequential algorithms, and almost ideal parallel speedup.

poster session 4 @ blue 6, poster session 9 @ blue 6, poster session 4, poster session 9

Representation learners that disentangle factors of variation have already proven to be important in addressing various real world concerns such as fairness and interpretability. Initially consisting of unsupervised models with independence assumptions, more recently, weak supervision and correlated features have been explored, but without a causal view of the generative process. In contrast, we work under the regime of a causal generative process where generative factors are either independent or can be potentially confounded by a set of observed or unobserved confounders. We present an analysis of disentangled representations through the notion of disentangled causal process. We motivate the need for new metrics and datasets to study causal disentanglement and propose two evaluation metrics and a dataset. We show that our metrics capture the desiderata of disentangled causal process. Finally we perform an empirical study on state of the art disentangled representation learners using our metrics and dataset to evaluate them from causal perspective.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

Pedestrian trajectory prediction has become an essential underpinning in various human-centric applications including but not limited to autonomous vehicles, intelligent surveillance system and social robotics. Previous research endeavors mainly focus on single camera trajectory prediction (SCTP), while the problem of multi-camera trajectory prediction (MCTP) is often overly simplified into predicting presence in the next camera. This paper addresses MCTP from a more realistic yet challenging perspective, by redefining the task as a joint estimation of both future destination and possible trajectory. As such, two major efforts are devoted to facilitating related research and advancing modeling techniques. Firstly, we establish a comprehensive multi-camera Scenes Pedestrian Trajectory Dataset (mcScenes), which is collected from a real-world multi-camera space combined with thorough human interaction annotations and carefully designed evaluation metrics. Secondly, we propose a novel joint prediction framework, namely HM3GAT, for the MCTP task by building a tailored network architecture. The core idea behind HM3GAT is a fusion of topological and trajectory information that are mutually beneficial to the prediction of each task, achieved by deeply customized networks. The proposed framework is comprehensively evaluated on the mcScenes dataset with multiple ablation experiments. Status-of-the-art SCTP models are adopted as baselines to further validate the advantages of our method in terms of both information fusion and technical improvement. The mcScenes dataset, the HM3GAT, and alternative models are made publicly available for interested readers.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

Description logic ontologies serve to model classifications and structural relationships, and to represent and reason about domain knowledge. Modeling the basic classification of abstract algebraic structures as an ontology demonstrates the difficulties presented by their logical semantics, and shed light on the limitations to accurately model further topics in algebra and related mathematical domains.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

In the practical application of restoring low-resolution gray-scale images, we generally need to run three separate processes of image colorization, super-resolution, and dows-sampling operation for the target device. However, this pipeline is redundant and inefficient for the independent processes, and some inner features could have been shared. Therefore, we present an efficient paradigm to perform \textit{\textbf{S}}imultaneously Image \textit{\textbf{C}}olorization and \textit{\textbf{S}}uper-resolution (SCS) and propose an end-to-end SCSNet to achieve this goal. The proposed method consists of two parts: colorization branch for learning color information that employs the proposed plug-and-play \emph{Pyramid Valve Cross Attention} (PVCAttn) module to aggregate feature maps between source and reference images; and super-resolution branch for integrating color and texture information to predict target images, which uses the designed \emph{Continuous Pixel Mapping} (CPM) module to predict high-resolution images at continuous magnification. Furthermore, our SCSNet supports both automatic and referential modes that is more flexible for practical application. Abundant experiments demonstrate the superiority of our method for generating authentic images over state-of-the-art methods, \eg, averagely decreasing FID by 1.8$\downarrow$ and 5.1 $\downarrow$ compared with current best scores for automatic and referential modes, respectively, while owning fewer parameters (more than $\times$2$\downarrow$) and faster running speed (more than $\times$3$\uparrow$).

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

Effectively integrating knowledge into end-to-end task-oriented dialog systems remains a challenge. It typically requires incorporation of an external knowledge base (KB) and capture of the intrinsic semantics of the dialog history. Recent research shows promising results by using Sequence-to-Sequence models, Memory Networks, and even Graph Convolutional Networks. However, current state-of-the-art models are less effective at integrating dialog history and KB into task-oriented dialog systems in the following ways: \textbf{1}. The KB representation is not fully context-aware. The dynamic interaction between the dialog history and KB is seldom explored. \textbf{2}. Both the sequential and structural information in the dialog history can contribute to capturing the dialog semantics, but they are not studied concurrently. In this paper, we propose a novel Graph Memory Network (GMN) based Seq2Seq model, GraphMemDialog, to effectively learn the inherent structural information hidden in dialog history, and to model the dynamic interaction between dialog history and KBs. We adopt a modified graph attention network to learn the rich structural representation of the dialog history, whereas the context-aware representation of KB entities are learnt by our novel GMN. To fully exploit this dynamic interaction, we design a learnable memory controller coupled with external KB entity memories to recurrently incorporate dialog history context into KB entities through a multi-hop reasoning mechanism. Experiments on three public datasets show that our GraphMemDialog model achieves state-of-the-art performance and outperforms strong baselines by a large margin, especially on datatests with more complicated KB information.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

In an out-of-distribution (OOD) detection problem, samples of known classes (also called in-distribution classes) are used to train a special classifier. In testing, the classifier can (1) classify the test samples of known classes to their respective classes and also (2) detect samples that do not belong to any of the known classes (i.e., they belong to some unknown or OOD classes). This paper studies the problem of zero-shot out-of-distribution (OOD) detection, which still performs the same two tasks in testing but has no training except using the given known class names. This paper proposes a novel and yet simple method (called ZOC) to solve the problem. ZOC builds on top of the recent advances in zero-shot classification through multi-modal representation learning. It first extends the pre-trained language-vision model CLIP by training a text-based image description generator on top of CLIP. In testing, it uses the extended model to generate candidate unknown class names for each test sample and computes a confidence score based on both the known class names and candidate unknown class names for zero-shot OOD detection. Experimental results on 5 benchmark datasets for OOD detection demonstrate that ZOC outperforms the baselines by a large margin.

poster session 4 @ red 4, poster session 11 @ red 4, poster session 4, poster session 11

Interactive AI (IAI) systems are increasingly popular as the human-centered AI design paradigm is gaining strong traction. However, evaluating IAI systems, a key step in building such systems, is particularly challenging, as their output highly depends on the performed user actions. Developers often have to rely on limited and mostly qualitative data from ad-hoc user testing to assess and improve their systems. In this paper, we present InteractEva; a systematic evaluation framework for IAI systems. We also describe how we have applied InteractEva to evaluate a commercial IAI system, leading to both quality improvements and better data-driven design decisions.

poster session 4 @ red 6, poster session 8 @ red 6, poster session 4, poster session 8

Convolutional neural network (CNN) based image segmentation has been widely used in analyzing medical images and benefited many real-world disease diagnosis applications. However, existing advanced CNN-based medical image segmentation models usually contain numerous parameters that require massive computation and memory, limiting the applicability of these models in the data-constrained or hardware-constrained environments. By leveraging the recently proposed neural architecture search (NAS), this paper presents a novel approach, dubbed Thrifty NAS, to design computation and memory-efficient models for medical image segmentation automatically. The searched models by Thrifty NAS are with much fewer parameters while retaining competitive performance. More specifically, we design a micro level space for cell structure search and a macro level cell path for better network structure modeling. Extensive experimental results in different medical image datasets verify the effectiveness of the proposed method with competitive segmentation performance, especially with minuscule neural architecture model size, i.e., 0.61M that is superior to U-Net (7.76 M) and UNet++ (9.04 M).

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

Knowledge-based visual question answering (VQA) involves answering questions that require external knowledge not present in the image. Existing methods first retrieve knowledge from external resources, then reason over the selected knowledge, the input image, and question for answer prediction. However, this two-step approach could lead to mismatches that potentially limit the VQA performance. For example, the retrieved knowledge might be noisy and irrelevant to the question, and the re-embedded knowledge features during reasoning might deviate from their original meanings in the knowledge base (KB). To address this challenge, we propose PICa, a simple yet effective method that Prompts GPT3 via the use of Image Captions, for knowledge-based VQA. Inspired by GPT-3’s power in knowledge retrieval and question answering, instead of using structured KBs as in previous work, we treat GPT-3 as an implicit and unstructured KB that can jointly acquire and process relevant knowledge. Specifically, we first convert the image into captions (or tags) that GPT-3 can understand, then adapt GPT-3 to solve the VQA task in a few-shot manner by just providing a few in-context VQA examples. We further boost performance by carefully investigating: (i) what text formats best describe the image content, and (ii) how in-context examples can be better selected and used. PICa unlocks the first use of GPT-3 for multimodal tasks. By using only 16 examples, PICa surpasses the supervised state of the art by an absolute +8.6 points on the OK-VQA dataset. We also benchmark PICa on VQAv2, where PICa also shows a decent few-shot performance.

poster session 4 @ red 2, poster session 11 @ red 2, oral session 11 @ red 2, poster session 4, poster session 11, oral session 11

Knowledge compilation is an alternative solution to address demanding reasoning tasks with high complexity via converting knowledge bases into a suitable target language. Interestingly, the notion of logical separability, proposed by Levesque, offers a general explanation for the tractability of clausal entailment for two remarkable languages: decomposable negation normal form and prime implicates. It is interesting to explore what role logical separability on earth plays in problem tractability. In this paper, we apply the notion of logical separability in three reasoning problems: satisfiability check (CO), clausal entailment check (CE) and model counting (CT), contributing to three corresponding polytime procedures. We provide three logical separability based properties: CO-logical separability, CE-logical separability and CT-logical separability. We then identify three novel normal forms: CO-LSNNF, CE-LSNNF and CT-LSNNF based on the above properties. Besides, we show that every normal form is the necessary and sufficient condition under which the corresponding procedure is correct. We finally integrate the above four normal forms into the knowledge compilation map.

poster session 4 @ blue 5, poster session 8 @ blue 5, oral session 8 @ blue 5, poster session 4, poster session 8, oral session 8

In this paper, we study the problem of eliciting preferences of agents in the house allocation model. For this we build on a recent model of Hosseini et al.[AAAI’21] and focus on the task of eliciting preferences to find matchings which are necessarily optimal, i.e., optimal under all possible completions of the elicited preferences. In particular, we follow the approach of Hosseini et al. and investigate the elicitation of necessarily Pareto optimal (NPO) and necessarily rank-maximal (NRM) matchings. Most importantly, we answer their open question and give an online algorithm for eliciting an NRM matching in the next-best query model which is 3/2-competitive,i.e., it takes at most 3/2 as many queries as an optimal algorithm. Besides this, we extend this field of research by introducing two new natural models of elicitation and by studying both the complexity of determining whether a necessarily optimal matching exists in them, and by giving online algorithms for these models.

poster session 4 @ blue 6, poster session 9 @ blue 6, oral session 4 @ blue 6, poster session 4, poster session 9, oral session 4

We analyze the assignment of passengers in a shared ride, which considers the social relationship among the passengers. Namely, there is a fixed number of passengers in each vehicle, and the goal is to recommend an assignment of the passengers such that the number of friendship relations is maximized. We show that the problem is computationally hard, and we provide an approximation algorithm.

poster session 4 @ blue 6, poster session 9 @ blue 6, poster session 4, poster session 9

This paper studies defender patrol deception in general Stackelberg security games (SSGs), where a defender attempts to alter the attacker's perception of the defender's patrolling intensity so as to influence the attacker's decision making. We are interested in understanding the complexity and effectiveness of optimal defender deception under different attacker behavior models. Specifically, we consider three different attacker strategies of response (to the defender's deception) with increasing sophistication, and design efficient polynomial-time algorithms to compute the equilibrium for each. Moreover, we prove formal separation for the effectiveness of patrol deception when facing an attacker of increasing sophistication, until it becomes even harmful to the defender when facing the most intelligent attacker we consider. Our results shed light on when and how deception should be used in SSGs. We conduct extensive experiments to illustrate our theoretical results in various game settings.

poster session 4 @ blue 6, poster session 9 @ blue 6, poster session 4, poster session 9

In a large domain of classification problems for real applications, like human activity recognition, separable spaces between groups of concepts are easier to learn than each concept alone. This is because the search space biases required to separate groups of classes (or concepts) are more relevant than the ones needed to separate classes individually. For example, it is easier to learn the activities related to the body movements group (e.g. running, walking, etc.) versus "on-wheels" activities group (e.g. bicycling, driving a car, etc.), before learning more specific classes inside each of these groups. Despite the obvious interest of this approach, our theoretical analysis shows a high complexity for finding an exact solution. We propose in this paper an original approach based on the association of clustering and classification approaches to overcome this limitation. We propose a better approach to learn the concepts by grouping classes recursively rather than learning them class by class. We introduce an effective greedy algorithm and two theoretical measures (namely cohesion and dispersion) to evaluate the connection between the clusters and the classes. Extensive experiments on the SHL dataset show that our approach improves classification performances while reducing the number of instances used to learn each concept.

poster session 4 @ red 4, poster session 11 @ red 4, oral session 11 @ red 4, poster session 4, poster session 11, oral session 11

Quantum computers can theoretically have significant acceleration over classical computers; but, the near-future era of quantum computing is limited due to small number of qubits that are also error prone. QUILT is a framework for performing multi-class classification task designed to work effectively on current error-prone quantum computers. QUILT is evaluated with real quantum machines as well as with projected noise levels as quantum machines become more noise free. QUILT demonstrates up to 85% multi-class classification accuracy with the MNIST dataset on a five-qubit system.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

Multi-label few-shot image classification (ML-FSIC) is the task of assigning descriptive labels to previously unseen images, based on a small number of training examples. A key feature of the multi-label setting is that images often have multiple labels, which typically refer to different regions of the image. When estimating prototypes, in a metric-based setting, it is thus important to determine which regions are relevant for which labels, but the limited amount of training data makes this highly challenging. As a solution, in this paper we propose to use word embeddings as a form of prior knowledge about the meaning of the labels. In particular, visual prototypes are obtained by aggregating the local feature maps of the support images, using an attention mechanism that relies on the label embeddings. As an important advantage, our model can infer prototypes for unseen labels without the need for fine-tuning any model parameters, which demonstrates its strong generalization abilities. Experiments on COCO and PASCAL VOC furthermore show that our model substantially improves the current state-of-the-art.

poster session 4 @ red 2, poster session 11 @ red 2, oral session 11 @ red 2, poster session 4, poster session 11, oral session 11

We consider the problem of reducing the false discovery rate in multiple high-dimensional interventional datasets under unknown intervention targets when the time efficiency is acceptable. Traditional algorithms merged directly multiple causal DAGs learned, which ignores the contradictions of different datasets, leading to lots of contradictory directions of edges. For reducing the contradictory information, we propose a new algorithm, which first learns an intervention Markov equivalence class (I-MEC) before merging multiple causal DAGs. It utilizes the full power of the constraints available in interventional data and combines ideas from local learning, intervention, and search-and-score techniques in a principled and effective way in different intervention experiments. Specifically, local learning on multiple datasets is used to build a causal skeleton. Perfect intervention destroys some possible triangles and obtains more V-structures, getting a theoretically correct I-MEC. Search and scoring techniques based on the learned I-MEC further identify the remaining unoriented edges. Experiments on benchmark Bayesian networks with the number of variables from 20 to 724 validate that the effectiveness of our algorithm in reducing the false discovery rate in high-dimensional interventional data.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

Commonsense question answering (CQA) aims to test if models can answer questions regarding commonsense knowledge that everyone knows. Prior works that incorporate external knowledge bases have shown promising results, but knowledge bases are expensive to construct and are often limited to a fixed set of relations. In this paper, we instead focus on better utilizing the \textit{implicit knowledge} stored in pre-trained language models. While researchers have found that the knowledge embedded in pre-trained language models can be extracted by having them fill in the blanks of carefully designed prompts for relation extraction and text classification, it remains unclear if we can adopt this paradigm in CQA where the inputs and outputs take much more flexible forms. To this end, we investigate four translation methods that can translate natural questions into cloze-style sentences to better solicit commonsense knowledge from language models, including a syntactic-based model, an unsupervised neural model, and two supervised neural models. In addition, to combine the different translation methods, we propose to encourage consistency among model predictions on different translated questions with unlabeled data. We demonstrate the effectiveness of our methods on three CQA datasets in zero-shot settings. We show that our methods are complementary to a knowledge base improved model, and combining them can lead to state-of-the-art zero-shot performance. Analyses also reveal distinct characteristics of the different cloze translation methods and provide insights on why combining them can lead to great improvements.

poster session 4 @ red 5, poster session 11 @ red 5, oral session 4 @ red 5, poster session 4, poster session 11, oral session 4

Sampling of chordal graphs and various types of acyclic orientations over chordal graphs plays a central role in several AI applications such as causal structure learning. For a given undirected graph, an acyclic orientation is an assignment of directions to all of its edges which makes the resulting directed graph cycle-free. Sampling is often closely related to the corresponding counting problem. Counting of acyclic orientations of a given chordal graph can be done in polynomial time, but the previously known techniques do not seem to lead to a corresponding (efficient) sampler. In this work, we propose a dynamic programming framework which yields a counter and a uniform sampler, both of which run in (essentially) linear time. An interesting feature of our sampler is that it is a stand-alone algorithm that, unlike other DP-based samplers, does not need any preprocessing which determines the corresponding counts.

poster session 4 @ blue 6, poster session 9 @ blue 6, poster session 4, poster session 9

This paper presents SimCTC, a simple contrastive learning (CL) framework that greatly advances the state-of-the-art text clustering models. In SimCTC, a pre-trained BERT model first maps the input sequence to the representation space, which is then followed by three different loss function heads: Clustering head, Instance-CL head and Cluster-CL head. Experimental results on multiple benchmark datasets demonstrate that SimCTC remarkably outperforms 6 competitive text clustering methods with 1%-6% improvement on Accuracy (ACC) and 1%-4% improvement on Normalized Mutual Information (NMI). Moreover, our results also show that the clustering performance can be further improved by setting an appropriate number of clusters in the cluster-level objective.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

Classical matrix completion methods focus on data with stationary latent structure and hence are not effective in missing value imputation when the latent structure changes with time. This paper proposes a dynamic nonlinear matrix completion (D-NLMC) method, which is able to recover the missing values of streaming data when the low-dimensional nonlinear latent structure of the data changes with time. We provide an efficient method to update the nonlinear model dynamically. It incorporates the information of the new data and remove the information of the earlier data recursively. We show that the missing data can be estimated if the change of latent structure is slow enough. Different from existing online or adaptive low-rank matrix completion methods, our method does not require the local low-rank assumption and is able to adaptively recover high-rank matrices with low-dimensional latent structures. Note that existing high-rank matrix completion methods have high-computational costs and are not applicable to streaming data with varying latent structures, which fortunately can be handled by our D-NLMC efficiently and accurately. Numerical results show that D-NLMC is more effective than the baselines.

poster session 4 @ red 4, poster session 11 @ red 4, poster session 4, poster session 11

Deep vision models are successfully employed in many applications. However, recent studies have shown that they are vulnerable to adversarial sample attacks. Existing adversarial defense methods are either limited to specific types of attacks or are too complex to be applied to practical vision models. More importantly, these methods rely on techniques that are not interpretable to humans. In this work, we argue that an effective defense method should produce the decision that explains why the system is attacked with a representation that is easily readable by a human user, e.g. a logic formalism. To this end, we propose logic adversarial defense (LogicDef), a defense framework that utilizes the scene graph of the image to provide a contextual structure for detecting and explaining object misclassification. Our framework first mines inductive logic rules from the graph that predict the objects, and then uses these rules to construct a defense model that alerts the users when the vision model violates the rules. The defense model is interpretable and its robustness can be further enhanced by incorporating commonsense knowledge from ConceptNet. Moreover, we propose a curriculum learning of the defense model based on object taxonomy which yields additional improvements in performance.

poster session 4 @ red 2, poster session 11 @ red 2, oral session 4 @ red 2, poster session 4, poster session 11, oral session 4

Existing studies on salient object detection (SOD) focus on extracting distinct objects with edge features and aggregating multi-level features to improve SOD performance. However, both performance gain and computational efficiency cannot be achieved, which has motivated us to study the inefficiencies in existing encoder-decoder structures to avoid this trade-off. We propose TRACER which excludes multi-decoder structures and minimizes the learning parameters usage by employing attention guided tracing modules (ATMs), as shown in Fig. 1.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

Learning from Crowds (LFC) seeks to induce a high-quality classifier from training instances, which are linked to a range of possible noisy annotations from crowdsourcing workers under their various levels of skills and their own preconditions. Recent studies on LFC focus on designing new methods to improve the performance of the classifier trained from crowdsourced labeled data. To this day, however, there remain under-explored security aspects of LFC systems. In this work, we seek to bridge this gap. We first show that LFC models are vulnerable to adversarial examples---small changes to input data can cause classifiers to make prediction mistakes. Second, we propose an approach, A-LFC for training a robust classifier from crowdsourced labeled data. Our empirical results on three real-world datasets show that the proposed approach can substantially improve the performance of the trained classifier even with the existence of adversarial examples. On average, A-LFC has 10.05% and 11.34% higher test robustness than the state-of-the-art in the white-box and black-box attack settings, respectively.

poster session 4 @ red 1, poster session 7 @ red 1, oral session 7 @ red 1, poster session 4, poster session 7, oral session 7

The mathematical formalization of a neurological mechanism in the olfactory circuit of a fruit-fly as a locality sensitive hash (Flyhash) and bloom filter (FBF) has been recently proposed and ``reprogrammed'' for various machine learning tasks such as similarity search, outlier detection and text embeddings. We propose a novel reprogramming of this hash and bloom filter to emulate the canonical nearest neighbor classifier (NNC) in the challenging Federated Learning (FL) setup where training and test data are spread across parties and no data can leave their respective parties. Specifically, we utilize Flyhash and FBF to create the FlyNN classifier, and theoretically establish conditions where FlyNN matches NNC. We show how FlyNN is trained {\em exactly} in a FL setup with low communication overhead to produce FlyNNFL, and how it can be differentially private. Empirically, we demonstrate that (i) FlyNN matches NNC accuracy across 70 OpenML datasets, (ii) FlyNNFL training is highly scalable with low communication overhead, providing up to $8\times$ speedup with $16$ parties.

poster session 4 @ red 4, poster session 11 @ red 4, oral session 11 @ red 4, poster session 4, poster session 11, oral session 11

This paper presents a novel collaborative generative modeling (CGM) framework that incentivizes collaboration among self-interested parties to contribute data to a pool for training a generative model (e.g., GAN), from which synthetic data are drawn and distributed to the parties as rewards commensurate to their contributions. Distributing synthetic data as rewards (instead of trained models or money) offers task- and model-agnostic benefits for downstream learning tasks and is less likely to violate data privacy regulation. To realize the framework, we firstly propose a data valuation function using maximum mean discrepancy (MMD) that values data based on its quantity and quality in terms of its closeness to the true data distribution and provide theoretical results guiding the kernel choice in our MMD-based data valuation function. Then, we formulate the reward scheme as a linear optimization problem that when solved, guarantees certain incentives such as fairness in the CGM framework. We devise a weighted sampling algorithm for generating synthetic data to be distributed to each party as reward such that the value of its data and the synthetic data combined matches its assigned reward value by the reward scheme. We empirically show using simulated and real-world datasets that the parties' synthetic data rewards are commensurate to their contributions.

poster session 4 @ red 3, poster session 8 @ red 3, oral session 4 @ red 3, poster session 4, poster session 8, oral session 4

Model transparency is a prerequisite in many domains and an increasingly popular area in machine learning research. In the medical domain, for instance, unveiling the mechanisms behind a disease often has higher priority than the diagnostic itself since it might dictate or guide potential treatments and research directions. One of the most popular approaches to explain model global predictions is the permutation importance where the performance on permuted data is benchmarked against the baseline. However, this method and other related approaches will undervalue the importance of a feature in the presence of covariates since these cover part of its provided information. To address this issue, we propose Covered Information Disentanglement CID, a framework that considers all feature information overlap to correct the values provided by permutation importance. We further show how to compute \CID efficiently when coupled with Markov random fields. We demonstrate its efficacy in adjusting permutation importance first on a controlled toy dataset and discuss its effect on real-world medical data.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

The success of deep neural nets heavily relies on their ability to encode complex relations between their input and their output. While this property serves to fit the training data well, it also obscures the mechanism that drives prediction. This study aims to reveal hidden concepts by employing an intervention mechanism that shifts the predicted class based on discrete variational autoencoders. An explanatory model then visualizes the encoded information from any hidden layer and its corresponding intervened representation. By the assessment of differences between the original representation and the intervened representation, one can determine the concepts that can alter the class, hence providing interpretability. We demonstrate the effectiveness of our approach on CelebA, where we show various visualizations for bias in the data and suggest different interventions to reveal and change bias.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

This demonstration paper introduces the Smart Out-of-Home Advertising Platform (SOAP), which leverages Geographic Information Systems (GIS) data and state-of-the-art Artificial Intelligence (AI) approaches to provide: (i) a documented, data-informed pricing model for billboards, which can be used to justify billboard prices to advertisers; and (ii) a set of non-dominated solutions (each corresponding to a different allocation of billboards to a given campaign) that explores the trade-offs between multiple conflicting objectives (e.g., cost and coverage). To the best of our knowledge, SOAP is the first to tackle such challenges in the context of Multi Objective Optimization (MOO).

poster session 4 @ blue 1, poster session 11 @ blue 1, poster session 4, poster session 11

Dialogue is an essential part of human communication and cooperation. Existing research mainly focuses on short dialogue scenarios in a one-on-one fashion. However, multi-person interactions in the real world, such as meetings or interviews, are frequently over a few thousand words. There is still a lack of corresponding research and powerful tools to understand and process such long dialogues. Therefore, in this work, we present a pre-training framework for long dialogue understanding and summarization. Considering the nature of long conversations, we propose a window-based denoising approach for generative pre-training. For a dialogue, it corrupts a window of text with dialogue-inspired noise, and guides the model to reconstruct this window based on the content of the remaining conversation. Furthermore, to process longer input, we augment the model with sparse attention which is combined with conventional attention in a hybrid manner. We conduct extensive experiments on five datasets of long dialogues, covering tasks of dialogue summarization, abstractive question answering and topic segmentation. Experimentally, we show that our pre-trained model \DialogLM significantly surpasses the state-of-the-art models across datasets and tasks.

poster session 4 @ red 5, poster session 11 @ red 5, oral session 11 @ red 5, poster session 4, poster session 11, oral session 11

Recent work has shown that Binarized Neural Networks (BNNs) are able to greatly reduce computational costs and memory footprints, facilitating model deployment on resource-constrained devices. However, in comparison to their full-precision counterparts, BNNs suffer from severe accuracy degradation. Research aiming to reduce this accuracy gap has thus far largely focused on specific network architectures with few or no 1 × 1 convolutional layers, for which standard binarization methods do not work well. Because 1 × 1 convolutions are common in the design of modern architectures (e.g. GoogleNet, ResNet, DenseNet), it is crucial to develop a method to binarize them effectively for BNNs to be more widely adopted. In this work, we propose an “Elastic-Link” (EL) module to enrich information flow within a BNN by adaptively adding real-valued input features to the subsequent convolutional output features. The proposed EL module is easily implemented and can be used in conjunction with other methods for BNNs. We demonstrate that adding EL to BNNs produces a significant improvement on the challenging large-scale ImageNet dataset. For example, we raise the top-1 accuracy of binarized ResNet26 from 57.9% to 64.0%. EL also aids con-vergence in the training of binarized MobileNet, for which a top-1 accuracy of 56.4% is achieved. Finally, with the integration of ReActNet, it yields a new state-of-the-art result of 71.9% top-1 accuracy.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

In recent years, companies in the Architecture, Engineering, and Construction (AEC) industry have started exploring how artificial intelligence (AI) can reduce time-consuming and repetitive tasks. One use case that can benefit from the adoption of AI is the determination of quantities in floor plans. This information is required for several planning and construction steps. Currently, the task requires companies to invest a significant amount of manual effort. Either digital floor plans are not available for existing buildings, or the formats cannot be processed due to lack of standardization. In this paper, we therefore propose a human-in-the-loop approach for the detection and classification of symbols in floor plans. The developed system calculates a measure of uncertainty for each detected symbol which is used to acquire the knowledge of human experts for those symbols that are difficult to classify. We evaluate our approach with a real-world dataset provided by an industry partner and find that the selective acquisition of human expert knowledge enhances the model’s performance by up to 10.5%—resulting in an overall prediction accuracy of 92.1% on average. We further design a pipeline for the generation of synthetic training data that allows the systems to be adapted to new construction projects with minimal manual effort. Overall, our work supports professionals in the AEC industry on their journey to the data-driven generation of business value.

poster session 4 @ red 6, poster session 8 @ red 6, poster session 4, poster session 8

Recently, neural networks based purely on self-attention, such as the Vision Transformer (ViT), have been shown to outperform deep learning models constructed with convolutional neural networks (CNNs) on various vision tasks, thus extending the success of Transformers, which were originally developed for language processing, to the vision domain. A recent study showed that a similar methodology can also be applied to the audio domain. Specifically, the Audio Spectrogram Transformer (AST) achieves state-of-the-art results on various audio classification benchmarks. However, pure Transformer models tend to require more training data compared to CNNs, and the success of the AST relies on supervised pretraining that requires a large amount of labeled data and a complex training pipeline, thus limiting the practical usage of AST. This paper focuses on audio and speech classification, and aims to reduce the need for large amounts of labeled data for the AST by leveraging self-supervised learning using unlabeled data. Specifically, we propose to pretrain the AST model with joint discriminative and generative masked spectrogram patch modeling (MSPM) using unlabeled audio from AudioSet and Librispeech. We evaluate our pretrained models on both audio and speech classification tasks including audio event classification, keyword spotting, emotion recognition, and speaker identification. The proposed self-supervised framework significantly boosts AST performance on all tasks, with an average improvement of 60.9%, leading to similar or even better results than a supervised pretrained AST. To the best of our knowledge, it is the first patch-based self-supervised learning framework in the audio and speech domain, and also the first self-supervised learning framework for AST.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

Deep learning excels at learning task information from large amounts of data, but struggles with learning from declarative high-level knowledge that can be more succinctly expressed directly. In this work, we introduce PYLON, a neuro-symbolic training framework that builds on PyTorch to augment procedurally trained models with declaratively specified knowledge. PYLON lets users programmatically specify constraints as Python functions and compiles them into a differentiable loss, thus training predictive models that fit the data whilst satisfying the specified constraints. PYLON includes both exact as well as approximate compilers to efficiently compute the loss, employing fuzzy logic, sampling methods, and circuits, ensuring scalability even to complex models and constraints. Crucially, a guiding principle in designing PYLON is the ease with which any existing deep learning codebase can be extended to learn from constraints in a few lines code: a function that expresses the constraint, and a single line to compile it into a loss. Our demo comprises of models in NLP, computer vision, logical games, and knowledge graphs that can be interactively trained using constraints as supervision.

poster session 4 @ blue 1, poster session 11 @ blue 1, poster session 4, poster session 11

Existing research in the field of automated negotiation considers a negotiation architecture in which some of the negotiation components are designed separately by reinforcement learning (RL), but comprehensive negotiation strategy design has not been achieved. In this study, we formulated an RL model based on a Markov decision process (MDP) for bilateral multi-issue negotiations. We propose a versatile negotiating agent that can effectively learn various negotiation strategies and domains through comprehensive strategies using deep RL. We show that the proposed method can achieve the same or better utility than existing negotiation agents.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

Pan-sharpening in remote sensing imaging systems refers to generate a high-resolution multispectral image from a high-resolution panchromatic (PAN) image and its corresponding low-resolution multispectral (MS) image. Over the recent years, convolution neural network (CNN)-based methods have demonstrated superior learning capability and dominated this field. However, due to the limitation of the convolution operator, long-range spatial features are often obtained imprecisely, leading to the degradation of overall performance. To this end, we propose a novel method by exploiting a customized transformer architecture for long-range dependencies modeling, and an information-lossless invertible neural module for effective feature fusion. Specifically, the customized transformer formulates the PAN and MS features as queries and keys to encourage joint feature learning across two modalities while the designed invertible neural module enables effective feature fusion to generate the expected pan-sharpened results. To the best of our knowledge, this is the first attempt to introduce transformer and invertible neural network on pan-sharpening. Extensive experiments over different satellite datasets demonstrate that our method outperforms the state-of-the-art algorithms both visually and quantitatively with fewer parameters and flops. Furthermore, the ablation experiments also prove the effectiveness of the proposed customized long-range transformer and the superiority of invertible neural feature fusion module for pan-sharpening.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

Keyphrase generation aims to summarize long documents with a collection of salient phrases. Deep neural models have demonstrated remarkable success in this task, capable of predicting keyphrases that are even absent from a document. However, such abstractiveness is acquired at the expense of a substantial amount of annotated data. In this paper, we present a novel method for keyphrase generation, AutoKeyGen, without the supervision of any human annotation. Motivated by the observation that an absent keyphrase in one document can appear in other places, in whole or in part, we first construct a phrase bank by pooling all phrases in a corpus. With this phrase bank, we then draw candidate absent keyphrases for each document through a partial matching process. To rank both types of candidates, we combine their lexical- and semantic-level similarities to the input document. Moreover, we utilize these top-ranked candidates as to train a deep generative model for more absent keyphrases. Extensive experiments demonstrate that AutoKeyGen outperforms all unsupervised baselines and can even beat strong supervised methods in certain cases.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

It is well-known that the search algorithms A* and Iterative Deepening A* (IDA*) can fail to solve state-space tasks optimally due to time and memory limits. The former typically fails in memory-restricted scenarios and the latter in time-restricted scenarios. Therefore, several algorithms were proposed to solve state-space tasks optimally using less memory than A* and less time than IDA*, such as A*+IDA*, a hybrid memory-restricted algorithm that combines A* and IDA*. In this paper, we present a hybrid memory-restricted algorithm that combines Partial Expansion A* (PEA*) and IDA*. This new algorithm has two phases, the same structure as the A*+IDA* algorithm. The first phase of PEA*+IDA* runs PEA* until it reaches a memory limit, and the second phase runs IDA* without duplicate detection on each node of PEA*'s Open. First, we present a model that shows how PEA*+IDA* can perform better than A*+IDA* although pure PEA* usually makes more expansions than pure A*. Later, we perform an experimental evaluation using three memory limits and show that compared to A*+IDA* on classical planning domains, PEA*+IDA* has higher coverage and expands fewer nodes. Finally, we experimentally analyze both algorithms and show that having higher F-limits and better priority-queue composition given by PEA* have a considerable impact on the performance of the algorithms.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

This paper addresses the problem of estimating causal effects when adjustment variables in the back-door or front-door criterion are partially observed. For such scenarios, we derive bounds on the causal effects by solving two non-linear optimization problems, and demonstrate that the bounds are sufficient. Using this optimization method, we propose a framework for dimensionality reduction that allows one to trade bias for estimation power, and demonstrate its performance using simulation studies.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

Magnetic Resonance Imaging (MRI) is a non-invasive imaging modality that is a cornerstone of diagnostic radiology. Clinical MRI scans capture a single image to highlight a single tissue property. The intensity difference between different regions of this image shows disease states that a radiologist can interpret. Magnetic Resonance Fingerprinting (MRF) is a recently proposed novel MRI technique. MRF allows the capture of multiple MR images in a single scan. This enables clinicians to analyze multiple tissue properties, potentially increasing the sensitivity of diagnosis and also allowing for the diagnosis of novel diseases. However, it is more challenging to analyze MRF images, because MRF produces much larger and noisier data than MRI. In this paper, we show how AI techniques can help solve this problem. Using a hybrid search strategy combining simulated annealing with pattern search, we show it is possible to tractably reconstruct multiple tissue properties from a single MRF image. This is a key step towards the deployment of MRF for radiological diagnosis.

poster session 4 @ red 6, poster session 9 @ red 6, poster session 4, poster session 9

We present a generative neural model for open and multi-turn dialog response generation that relies on a multi-dimension attention process to account for the semantic interdependence between the generated words and the conversational history, so as to identify all the words and utterances that influence each generated response. The performance of the model is evaluated on the wide scope DailyDialog corpus and a comparison is made with two other generative neural architectures, using several machine metrics. The results show that the proposed model improves the state of the art for generation accuracy, and its multi-dimension attention allows for a more detailed tracking of the influential words and utterances in the dialog history for response explainability by the dialog history.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

In black-box optimization problems, we aim to maximize an unknown objective function, where the function is only accessible through feedbacks of an evaluation or simulation oracle. In real-life, the feedbacks of such oracles are often noisy and available after some unknown delay that may depend on the computation time of the oracle. Additionally, if the exact evaluations are expensive but coarse approximations are available at a lower cost, the feedbacks can have multi-fidelity. In order to address this problem, we propose a generic extension of hierarchical optimistic tree search (HOO), called ProCrastinated Tree Search (PCTS), that flexibly accommodates a delay and noise-tolerant bandit algorithm. We provide a generic proof technique to quantify regret of PCTS under delayed, noisy, and multi-fidelity feedbacks. Specifically, we derive regret bounds of PCTS enabled with delayed-UCB1 (DUCB1) and delayed-UCB-V (DUCBV) algorithms. Given a horizon T, PCTS retains the regret bound of non-delayed HOO for expected delay of O(log T), and worsens by T^{(1-\alpha)/(d+2)} for expected delays of O(T^{1-\alpha}) for \alpha \in (0,1]. We experimentally validate on multiple synthetic functions and hyperparameter tuning problems that PCTS outperforms the state-of-the-art black-box optimization methods for feedbacks with different noise levels, delays, and fidelity.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

Accounts Payable (AP) is a resource-intensive business process in large enterprises for paying vendors within contractual payment deadlines for goods and services procured from them. There are multiple verifications before payment to the supplier/vendor. After the validations, the invoice flows through several steps such as vendor identification, line-item matching for Purchase order (PO) based invoices, Accounting Code identification for Non- Purchase order (Non-PO) based invoices, tax code identification, etc. Currently, each of these steps is mostly manual and cumbersome making it labor-intensive, error-prone, and requiring constant training of agents. Automatically processing these invoices for payment without any manual intervention is quite difficult. To tackle this challenge, we have developed an automated end-to-end invoice processing system using AI-based modules for multiple steps of the invoice processing pipeline. It can be configured to an individual client’s requirements with minimal effort. Currently, the system is deployed in production for two clients. It has successfully processed around ~80k invoices out of which 76% invoices were processed with low or no manual intervention.

poster session 4 @ red 6, poster session 8 @ red 6, poster session 4, poster session 8

Binary networks are extremely efficient as they use only two symbols to define the network: {+1, −1}. One can make the prior distribution of these symbols a design choice. The recent IR-Net of Qin et al. argues that imposing a Bernoulli distribution with equal priors (equal bit ratios) over the binary weights leads to maximum entropy and thus minimizes information loss. However, prior work cannot precisely control the binary weight distribution during training, and therefore cannot guarantee maximum entropy. Here, we show that quantizing using optimal transport can guarantee any bit ratio, including equal ratios. We investigate experimentally that equal bit ratios are indeed preferable and show that our method leads to optimization benefits. We show that our quantization method is effective when compared to state-of-the-art binarization methods, even when using binary weight pruning.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

Conventional static knowledge graphs model entities in relational data as nodes, connected by edges of specific relation types. However, information and knowledge evolve continuously, and temporal dynamics emerge, which are expected to influence future situations. In temporal knowledge graphs, time information is integrated into the graph by equipping each edge with a timestamp or a time range. Embedding-based methods have been introduced for link prediction on temporal knowledge graphs, but they mostly lack explainability and comprehensible reasoning chains. Particularly, they are usually not designed to deal with link forecasting -- event prediction involving future timestamps. We address the task of link forecasting on temporal knowledge graphs and introduce TLogic, an explainable framework that is based on temporal logical rules extracted via temporal random walks. We compare TLogic with state-of-the-art baselines on three benchmark datasets and show better overall performance while our method also provides explanations that preserve time consistency. Furthermore, in contrast to most state-of-the-art embedding-based methods, TLogic works well in the inductive setting where already learned rules are transferred to related datasets with a common vocabulary.

poster session 4 @ blue 2, poster session 7 @ blue 2, poster session 4, poster session 7

Unlike traditional supervised learning, in many settings only partial feedback is available. We may only observe outcomes for the chosen actions, but not the counterfactual outcomes associated with other alternatives. Such settings encompass a wide variety of applications including pricing, online marketing and precision medicine. A key challenge is that observational data are influenced by historical policies deployed in the system, yielding a biased data distribution. We approach this task as a domain adaptation problem and propose a self-training algorithm which imputes outcomes with \textit{categorical} values for \textit{finite} unseen actions in the observational data to simulate a randomized trial through pseudolabelling, which we refer to as Counterfactual Self-Training (CST). CST iteratively imputes pseudolabels and retrains the model. In addition, we show input consistency loss can further improve CST performance which is shown in recent theoretical analysis of pseudolabelling. We demonstrate the effectiveness of the proposed algorithms on both synthetic and real datasets.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

One of the most widespread human behavioral biases is the present bias -- the tendency to overestimate current costs by a bias factor. Kleinberg and Oren (2014) introduced an elegant graph-theoretical model of inconsistent planning capturing the behavior of a present-biased agent accomplishing a set of actions. The essential measure of the system introduced by Kleinberg and Oren is the cost of irrationality -- the ratio of the total cost of the actions performed by the present-biased agent to the optimal cost. This measure is vital for a task designer to estimate the aftermaths of human behavior related to time-inconsistent planning, including procrastination and abandonment. As we prove in this paper, the cost of irrationality is highly susceptible to the agent's choices when faced with a few possible actions of equal estimated costs. To address this issue, we propose a modification of Kleinberg-Oren's model of inconsistent planning. In our model, when an agent selects from several options of minimum prescribed cost, he uses a randomized procedure. We explore the algorithmic complexity of computing and estimating the cost of irrationality in the new model.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

Hawkes processes are a special class of temporal point processes which exhibit a natural notion of causality, as occurrence of events in the past may increase the probability of events in the future. Discovery of the underlying influence network among the dimensions of multi-dimensional temporal processes is of high importance in disciplines where a high-frequency data is to model, e.g. in financial data or in seismological data. This paper approaches the problem of learning Granger-causal network in multi-dimensional Hawkes processes. We formulate this problem as a model selection task in which we follow the minimum description length (MDL) principle. Moreover, we propose a general algorithm for MDL-based inference using a Monte-Carlo method and we use it for our causal discovery problem. We compare our algorithm with the state-of-the-art baseline methods on synthetic and real-world financial data. The synthetic experiments demonstrate superiority of our method in causal graph discovery compared to the baseline methods with respect to the size of the data. The results of experiments with the G-7 bonds price data are consistent with the experts’ knowledge.

poster session 4 @ blue 3, poster session 9 @ blue 3, oral session 9 @ blue 3, poster session 4, poster session 9, oral session 9

Sample average approximation (SAA), a popular method for tractably solving stochastic optimization problems, enjoys strong asymptotic performance guarantees in settings with independent training samples. However, these guarantees are not known to hold generally with dependent samples, such as in online learning with time series data or distributed computing with Markovian training samples. In this paper, we show that SAA remains tractable when the distribution of unknown parameters is only observable through dependent instances and still enjoys asymptotic consistency and finite sample guarantees. Specifically, we provide a rigorous probability error analysis to derive $1 - \beta$ confidence bounds for the out-of-sample performance of SAA estimators and show that these estimators are asymptotically consistent. We then, using monotone operator theory, study the performance of a class of stochastic first-order algorithms trained on a dependent source of data. We show that approximation error for these algorithms is bounded and concentrates around zero, and establish deviation bounds for iterates when the underlying stochastic process is $\phi$-mixing. The algorithms presented can be used to handle numerically inconvenient loss functions such as the sum of a smooth and non-smooth function or of non-smooth functions with constraints. To illustrate the usefulness of our results, we present several stochastic versions of popular algorithms such as stochastic proximal gradient descent (S-PGD), stochastic relaxed Peaceman--Rachford splitting algorithms (S-rPRS), and numerical experiment.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

RGB-NIR fusion is a promising method for low-light imaging. However, high-intensity noise in low-light images amplifies the effect of structure inconsistency between RGB-NIR images, which fails existing algorithms. To handle this, we propose a new RGB-NIR fusion algorithm called Dark Vision Net (DVN) with two technical novelties: Deep Structure and Deep Inconsistency Prior (DIP). The Deep Structure extracts clear structure details in deep multiscale feature space rather than raw input space, which is more robust to noisy inputs. Based on the deep structures from both RGB and NIR domains, we introduce the DIP to leverage the structure inconsistency to guide the fusion of RGB-NIR. Benefits from this, the proposed DVN obtains high-quality low-light images without the visual artifacts. We also propose a new dataset called Dark Vision Dataset (DVD), consisting of aligned RGB-NIR image pairs, as the first public RGB-NIR fusion benchmark. Quantitative and qualitative results on the proposed benchmark show that DVN significantly outperforms other comparison algorithms in PSNR and SSIM, especially in extremely low light conditions.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

We consider an application of combinatorial search to the optimization of topologies in series-parallel networks. We propose a recursive search over the space of decomposition trees, in which partial solutions are obtained by exploring k-way partitionings of expandable nodes. We present two complementary pruning techniques that bound the value of intermediate solutions from above and below, applying monotonic operations to the contents of unresolved leaves. We also develop a means to exploit the convexity of our objective function, so as to prevent the redundant recomputation of subcircuit configurations. Finally, we evaluate our approach on a parameterized benchmark suite of electrical circuits, demonstrating over an order of magnitude improvement in performance as compared to a baseline implementation.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

Understanding the black-box prediction for neural networks is challenging. To achieve this, early studies have designed influence function (IF) to measure the effect of removing a single training point on neural networks. However, the classic implicit Hessian-vector product (IHVP) method for calculating IF is fragile, and theoretical analysis of IF in the context of neural networks is still lacking. To this end, we utilize the neural tangent kernel (NTK) theory to calculate IF for the neural network trained with regularized mean-square loss, and prove that the approximation error can be arbitrarily small when the width is sufficiently large for two-layer ReLU networks. We analyze the error bound for the classic IHVP method in the over-parameterized regime to understand when and why it fails or not. In detail, our theoretical analysis reveals that (1) the accuracy of IHVP depends on the regularization term, and is pretty low under weak regularization; (2) the accuracy of IHVP has a significant correlation with the probability density of corresponding training points. We further borrow the theory from NTK to understand the IFs better, including quantifying the complexity for influential samples and depicting the variation of IFs during the training dynamics. Numerical experiments on real-world data confirm our theoretical results and demonstrate our findings.

poster session 4 @ blue 3, poster session 9 @ blue 3, oral session 9 @ blue 3, poster session 4, poster session 9, oral session 9

Emerging high-quality face restoration (FR) methods often utilize pre-trained GAN models (i.e., StyleGAN2) as GAN Prior. However, these methods usually struggle to balance realness and fidelity when facing various degradation levels. Besides, there is still a noticeable visual quality gap compared with pre-trained GAN models. In this paper, we propose a novel GAN Prior based degradation-aware feature interpolation network, dubbed Panini-Net, for FR tasks by explicitly learning the abstract representations to distinguish various degradations. Specifically, an unsupervised degradation representation learning (UDRL) strategy is first developed to extract degradation representations (DR) of the input degraded images. Then, a degradation-aware feature interpolation (DAFI) module is proposed to dynamically fuse the two types of informative features (i.e., features from input images and features from GAN Prior) with flexible adaption to various degradations based on DR. Ablation studies reveal the working mechanism of DAFI and its potential for editable FR. Extensive experiments demonstrate that our Panini-Net achieves state-of-the-art performance for multi-degradation face restoration and face super-resolution.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

Recently, deep learning techniques have been extensively studied for pansharpening, which aims to generate a high resolution multispectral (HRMS) image by fusing a low resolution multispectral (LRMS) image with a high resolution panchromatic (PAN) image. However, existing deep learning-based pansharpening methods directly learn the mapping from LRMS and PAN to HRMS. These network architectures always lack sufficient interpretability, which limits further performance improvements. To alleviate this issue, we propose a novel deep network for pansharpening by combining the model-based methodology with the deep learning method. Firstly, we build an observation model for pansharpening using the convolutional sparse coding (CSC) technique and design a proximal gradient algorithm to solve this model. Secondly, we unfold the iterative algorithm into a deep network, dubbed as Proximal PanNet, by learning the proximal operators using convolutional neural networks. Finally, all the learnable modules can be automatically learned in an end-to-end manner. Experimental results on some benchmark datasets show that our network performs better than other advanced methods both quantitatively and qualitatively.

poster session 4 @ red 3, poster session 8 @ red 3, oral session 8 @ red 3, poster session 4, poster session 8, oral session 8

Blackbox transfer attacks for image classifiers have been extensively studied in recent years. In contrast, little progress has been made on transfer attacks for object detectors. Object detectors take a holistic view of the image and the detection of one object (or lack thereof) often depends on other objects in the scene. This makes such detectors inherently context-aware and adversarial attacks in this space are more challenging than those targeting image classifiers. In this paper, we present a new approach to generate context-aware attacks for object detectors. We show that by using co-occurrence of objects and their relative locations and sizes as context information, we can successfully generate targeted mis-categorization attacks that achieve higher transfer success rates on blackbox object detectors than the state-of-the-art. We test our approach on a variety of object detectors with images from PASCAL VOC and MS COCO datasets and demonstrate up to $20$ percentage points improvement in performance compared to the other state-of-the-art methods.

poster session 4 @ red 2, poster session 11 @ red 2, poster session 4, poster session 11

We address the poor scalability of learning algorithms for orthogonal recurrent neural networks via the use of stochastic coordinate descent on the orthogonal group, leading to a cost per iteration that increases linearly with the number of recurrent states. This contrasts with the cubic dependency of typical feasible algorithms such as stochastic Riemannian gradient descent, which prohibits the use of big network architectures. Coordinate descent rotates successively two columns of the recurrent matrix. When the coordinate (i.e., indices of rotated columns) is selected uniformly at random at each iteration, we prove convergence of the algorithm under standard assumptions on the loss function, stepsize and minibatch noise. In addition, we numerically show that the Riemannian gradient has an approximately sparse structure. Leveraging this observation, we propose a variant of our proposed algorithm that relies on the Gauss-Southwell coordinate selection rule. Experiments on a benchmark recurrent neural network training problem show that the proposed approach is a very promising step towards the training of orthogonal recurrent neural networks with big architectures.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

In this paper, we propose a novel method to mine the commonsense knowledge shared between the video and text modalities for video-text retrieval, namely visual consensus modeling. Different from the existing works, which learn the video and text representations and their complicated relationships solely based on the pairwise video-text data, we make the first attempt to model the visual consensus by mining the visual concepts from videos and exploiting their co-occurrence patterns within the video and text modalities with no reliance on any additional concept annotations. Specifically, we build a shareable and learnable graph as the visual consensus, where the nodes denoting the mined visual concepts and the edges connecting the nodes representing the co-occurrence relationships between the visual concepts. Extensive experimental results on the public benchmark datasets demonstrate that our proposed method, with the ability to effectively model the visual consensus, achieves the state-of-the-art performances on the bidirectional video-text retrieval task. Our code is available at https://github.com/sqiangcao99/VCM.

poster session 4 @ red 2, poster session 11 @ red 2, poster session 4, poster session 11

Continuous-depth neural networks, such as the Neural Ordinary Differential Equations (ODEs), have aroused a great deal of interest from the communities of machine learning and data science in recent years, which bridge the connection between deep neural networks and dynamical systems. In this article, we introduce a new sort of continuous-depth neural network, called the Neural Piecewise-Constant Delay Differential Equations (PCDDEs). Here, unlike the recently proposed framework of the Neural Delay Differential Equations (DDEs), we transform the single delay into the piecewise-constant delay(s). The Neural PCDDEs with such a transformation, on one hand, inherit the strength of universal approximating capability in Neural DDEs. On the other hand, the Neural PCDDEs, leveraging the contributions of the information from the multiple previous time steps, further promote the modeling capability without augmenting the network dimension. With such a promotion, we show that the Neural PCDDEs do outperform the several existing continuous-depth neural frameworks on the one-dimensional piecewise-constant delay population dynamics and real-world datasets, including MNIST, CIFAR10, and SVHN.

poster session 4 @ red 4, poster session 11 @ red 4, poster session 4, poster session 11

Most economic reports suggest that almost half of the market value unlocked by artificial intelligence (AI) by the next decade (about 9 trillion USD per year) will be in marketing&sales. In particular, AI will allow the optimization of more and more intricate economic settings in which multiple different activities can be automated jointly. A relatively recent example is that one of ad auctions in which similar products or services are displayed together with their price, thus merging advertising and pricing in a unique website. This is the case, e.g., of Google Hotel Ads and TripAdvisor. More precisely, as in a classical ad auction, the ranking of the ads depends on the advertisers' bids, while, differently from classical ad auctions, the price is displayed together with the ad, so as to provide a direct comparison among the prices and thus dramatically affect the behavior of the users. This paper investigates how displaying prices and ads together conditions the properties of the main economic mechanisms such as VCG and GSP. Initially, we focus on the direct-revelation mechanism, showing that prices are chosen by the mechanisms once given the advertisers' reports. We also provide an efficient algorithm to compute the optimal allocation given the private information reported by the advertisers. Then, with both VCG and GSP payments, we show the inefficiency in terms of Price of Anarchy (PoA) and Stability (PoS) over the social welfare and mechanism's revenue when the advertisers choose the prices. The main results show that, with both VCG and GSP, PoS over the revenue may be unbounded even with two slots, while PoA over the social welfare may be as large as the number of slots. Finally, we show that, under some assumptions, simple modifications to VCG and GSP allow us to obtain a better PoS over the revenue.

poster session 4 @ blue 6, poster session 9 @ blue 6, oral session 4 @ blue 6, poster session 4, poster session 9, oral session 4

Causality plays a central role in reasoning about observations. In many cases, it might be useful to define the conditions under which a non-deterministic program can be called an actual cause of an effect in a setting where a sequence of pro- grams are executed one after another. There can be two perspectives, one where at least one execution of the program leads to the effect, and another where all executions do so. The former captures a ``weak'' notion of causation and is more general than the latter stronger notion. In this paper, we give the definition of weak potential causes. Our analysis is performed within the situation calculus basic action theories and we consider programs formulated in the logic programming language ConGolog. Within this setting, we show how one can utilize a recently developed abstraction framework to relate causes at various levels of abstraction. In particular, we prove that under some conditions, causes defined at an abstract and a concrete level can be related with each other in a kind of commutative diagram.

poster session 4 @ blue 5, poster session 8 @ blue 5, poster session 4, poster session 8

Recent research has demonstrated that Deep Neural Networks (DNNs) are vulnerable to adversarial patches which introduce perceptible but localized changes to the input. Nevertheless, existing approaches have focused on generating adversarial patches on images, their counterparts in videos have been less explored. Compared with images, attacking videos is much more challenging as it needs to consider not only spatial cues but also temporal cues. To close this gap, we introduce a novel adversarial attack in this paper, the bullet-screen comment (BSC) attack, which attacks video recognition models with BSCs. Specifically, adversarial BSCs are generated with a Reinforcement Learning (RL) framework, where the environment is set as the target model and the agent plays the role of selecting the position and transparency of each BSC. By continuously querying the target models and receiving feedback, the agent gradually adjusts its selection strategies in order to achieve a high fooling rate with non-overlapping BSCs. As BSCs can be regarded as a kind of meaningful patch, adding it to a clean video will not affect people’s understanding of the video content, nor will arouse people’s suspicion. We conduct extensive experiments to verify the effectiveness of the proposed method. On both UCF-101 and HMDB-51 datasets, our BSC attack method can achieve about 90\% fooling rate when attacking three mainstream video recognition models, while only occluding \textless 8\% areas in the video. Our code is available at \url{https://github.com/kay-ck/BSC-attack}.

poster session 4 @ red 2, poster session 11 @ red 2, oral session 11 @ red 2, poster session 4, poster session 11, oral session 11

Random forests have long been considered as powerful model ensembles in machine learning. By training multiple decision trees, whose diversity is fostered through data and feature subsampling, the resulting random forest can lead to more stable and reliable predictions than a single decision tree. This however comes at the cost of decreased interpretability: while decision trees are often easily interpretable, the predictions made by random forests are much more difficult to understand, as they involve a majority vote over multiple decision trees. In this paper, we examine different types of reasons that explain “why” an input instance is classified as positive or negative by a Boolean random forest. Notably, as an alternative to prime-implicant explanations taking the form of subset-minimal implicants of the random forest, we introduce majoritary reasons which are subset-minimal implicants of a strict majority of decision trees. For these abductive explanations, the tractability of the generation problem (finding one reason) and the optimization problem (finding one shortest reason) are investigated. Experiments conducted on various datasets reveal the existence of a trade-off between runtime complexity and sparsity. In practice, prime-implicant explanations - for which the identification problem is DP-complete - are slightly larger than majoritary reasons that can be generated using a simple linear-time greedy algorithm. They are also significantly larger than minimum-size majoritary reasons which can be approached using an anytime PARTIAL MAXSAT algorithm.

poster session 4 @ blue 3, poster session 9 @ blue 3, oral session 4 @ blue 3, poster session 4, poster session 9, oral session 4

Modern multi-agent robotic systems increasingly require scalable, robust and persistent Multi-Agent Path Finding (MAPF) with performance guarantees. While many MAPF solvers that provide some of these properties exist, none provides them all. To fill this need, we propose a new MAPF framework, the shard system. A shard system partitions the workspace into geographic regions, called shards, linked by a novel system of buffers. Agents are routed optimally within a shard by a local controller to local goals set by a global controller. The buffer system novelly allows shards to plan with perfect parallelism, providing scalability. A novel global controller algorithm can rapidly generate an inter-shard routing plan for thousands of agents while minimizing the traffic routed through any shard. A novel workspace partitioning algorithm produces shards small enough to replan rapidly. These innovations allow a shard system to adjust its routing plan in real time if an agent is delayed or assigned a new goal, enabling robust, persistent MAPF. A shard system's local optimality and optimized inter-shard routing bring the sum-of-costs of its solutions to single-shot MAPF problems to < 20-60% of optimal on a diversity of workspaces. Its scalability allows it to plan paths for 1000s of agents in seconds. If any of their goals change or move actions fails, a shard system can replan in under a second.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

The Cumulative constraint greatly contributes to the success of constraint programming at solving scheduling problems. The SoftCumulative, a version of the Cumulative where overloading the resource incurs a penalty is, however, less studied. We introduce a checker and a filtering algorithm for the SoftCumulative, which are inspired by the powerful energetic reasoning rule for the Cumulative. Both algorithms can be used with classic linear penalty function, but also with a quadratic penalty function, where the penalty of overloading the resource increases quadratically with the amount of the overload. We show that these algorithms are more general than existing algorithms and vastly outperform a decomposition of the SoftCumulative in practice.

poster session 4 @ blue 4, poster session 8 @ blue 4, oral session 8 @ blue 4, poster session 4, poster session 8, oral session 8

In livestock management, the decision of animal replacement requires an estimation of the lifetime profit of the animal based on multiple factors and operational conditions. In Dairy farms, this can be associated with the profit corresponding to milk production, health condition and herd management costs, which in turn may be a function of other factors including genetics and weather conditions. Estimating the profit of a cow can be expressed as a spatio-temporal problem where knowing the first batch of production (early-profit) can allow to predict the future batch of productions (late-profit). This problem can be addressed either by a univariate or multivariate time series forecasting. Several approaches have been designed for time series forecasting including Auto-Regressive approaches, Recurrent Neural Network including Long Short Term Memory (LSTM) method and a very deep stack of fully-connected layers. In this paper, we proposed a LSTM based approach coupled with attention and linear layers to better capture the dairy features. We compare the model, with three other architectures including NBEATs, ARIMA, MUMU-RNN using dairy production of 292181 dairy cows. The results highlight the performence of the proposed model of the compared architectures. They also show that a univariate NBEATs could perform better than the multi-variate approach there are compared to. We also highlight that such architecture could allow to predict late-profit with an error less than 3$ per month, opening the way of better resource management in the dairy industry.

poster session 4 @ blue 2, poster session 7 @ blue 2, poster session 4, poster session 7

In this paper, I pose a major challenge for AI researchers: to develop systems that learn in a human-like manner. I briefly review the history of machine learning, noting that early work made close contact with results from cognitive psychology but that this is no longer the case. I identify seven characteristics of human behavior that, if reproduced, would offer better ways to acquire expertise than statistical induction over massive training sets. I illustrate these points with two domains - mathematics and driving - where people are effective learners and review systems that address them. In closing, I suggest ways to encourage more research on human-like learning.

poster session 4 @ blue 1, poster session 11 @ blue 1, oral session 4 @ blue 1, poster session 4, poster session 11, oral session 4

Pattern mining is an essential part of knowledge discovery and data analytics. It is a powerful paradigm, especially when combined with constraint reasoning. In this paper, we present Seq2Pat, a constraint-based sequential pattern mining tool with a high-level declarative user interface. The library finds patterns that frequently occur in large sequence databases subject to constraints. We highlight key benefits that are desirable, especially in industrial settings where scalability, explainability, rapid experimentation, reusability, and reproducibility are of great interest. We then showcase an automated feature extraction process powered by Seq2Pat to discover high-level insights and boost downstream machine learning models for customer intent prediction.

poster session 4 @ red 6, poster session 8 @ red 6, poster session 4, poster session 8

Physical adversarial attacks in object detection have attracted increasing attention. However, most previous works focus on hiding the objects from the detector by generating an individual adversarial patch, which only covers the planar part of the vehicle's surface and fails to attack the detector in physical scenarios for multi-view, long-distance and partially occluded objects. To bridge the gap between digital attacks and physical attacks, we exploit the full 3D vehicle surface to propose a robust Full-coverage Camouflage Attack (FCA) to fool detectors. Specifically, we first try rendering the nonplanar camouflage texture over the full vehicle surface. To mimic the real-world environment conditions, we then introduce a transformation function to transfer the rendered camouflaged vehicle into a photo realistic scenario. Finally, we design an efficient loss function to optimize the camouflage texture. Experiments show that the full-coverage camouflage attack can not only outperform state-of-the-art methods under various test cases but also generalize to different environments, vehicles, and object detectors. The code of FCA will be available at: https://idrl-lab.github.io/Full-coverage-camouflage-adversarial-attack/.

poster session 4 @ red 2, poster session 11 @ red 2, poster session 4, poster session 11

In this paper, we propose a new Hessian inverse free Fully Single Loop Algorithm (FSLA) for bilevel optimization problems. Classic algorithms for bilevel optimization admit a double loop structure which is computationally expensive. Recently, several single loop algorithms have been proposed with optimizing the inner and outer variable alternatively. However, these algorithms not yet achieve fully single loop. As they overlook the loop needed to evaluate the hyper-gradient for a given inner and outer state. In order to develop a fully single loop algorithm, we first study the structure of the hyper-gradient and identify a general approximation formulation of hyper-gradient computation that encompasses several previous common approaches, \emph{e.g.} back-propagation through time, conjugate gradient, \emph{etc.} Based on this formulation, we introduce a new state variable to maintain the historical hyper-gradient information. Combining our new formulation with the alternative update of the inner and outer variables, we propose an efficient fully single loop algorithm. We theoretically show that the error generated by the new state can be bounded and our algorithm converges with the rate of $O(\epsilon^{-2})$. Finally, we verify the efficacy our algorithm empirically through multiple bilevel optimization based machine learning tasks.

poster session 4 @ red 4, poster session 11 @ red 4, poster session 4, poster session 11

Effective communication can improve coordination in cooperative multi-agent reinforcement learning (MARL). One popular communication scheme is exchanging agents' local observations or latent embeddings and using them to augment individual local policy input. Such a communication paradigm can reduce uncertainty for local decision-making and induce implicit coordination. However, it enlarges agents' local policy spaces and increases learning complexity, leading to poor coordination in complex settings. To handle this limitation, this paper proposes a novel framework named Multi-Agent Incentive Communication (MAIC) that allows each agent to learn to generate incentive messages and bias other agents' value functions directly, resulting in effective explicit coordination. Our method firstly learns targeted teammate models, with which each agent can anticipate the teammate's action selection and generate tailored messages to specific agents. We further introduce a novel regularization to leverage interaction sparsity and improve communication efficiency. MAIC is agnostic to specific MARL algorithms and can be flexibly integrated with different value function factorization methods. Empirical results demonstrate that our method significantly outperforms baselines and achieves excellent performance on multiple cooperative MARL tasks.

poster session 4 @ blue 6, poster session 9 @ blue 6, poster session 4, poster session 9

Transformer language models have made tremendous strides in natural language understanding tasks. However, the complexity of natural language makes it challenging to ascertain how accurately these models are tracking the world state underlying the text. Motivated by this issue, we consider the task of language modeling for the game of chess. Unlike natural language, chess notations describe a simple, constrained, and deterministic domain. Moreover, we observe that the appropriate choice of chess notation allows for directly probing the world state, without requiring any additional probing-related machinery. We find that: (a) With enough training data, transformer language models can learn to track pieces and predict legal moves with high accuracy when trained solely on move sequences. (b) For small training sets providing access to board state information during training can yield significant improvements. (c) The success of transformer language models is dependent on access to the entire game history i.e. “full attention”. Approximating this full attention results in a significant performance drop. We propose this testbed as a benchmark for future work on the development and analysis of transformer language models.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

As artificial intelligence and machine learning algorithms become increasingly prevalent in society, multiple stakeholders are calling for these algorithms to provide explanations. At the same time, these stakeholders, whether they be affected citizens, government regulators, domain experts, or system developers, have different explanation needs. To address these needs, in 2019, we created AI Explainability 360, an open source software toolkit featuring ten diverse and state-of-the-art explainability methods and two evaluation metrics. This paper examines the impact of the toolkit with several case studies, statistics, and community feedback. The different ways in which users have experienced AI Explainability 360 have resulted in multiple types of impact and improvements in multiple metrics, highlighted by the adoption of the toolkit by the independent LF AI & Data Foundation. The paper also describes the flexible design of the toolkit, examples of its use, and the significant educational material and documentation available to its users.

poster session 4 @ red 6, poster session 8 @ red 6, poster session 4, poster session 8

Multi-Agent Path Finding (MAPF) is the problem of finding a set of collision-free paths for a team of agents in a common environment. MAPF is NP-hard to solve optimally and, in some cases, also bounded suboptimally. It is thus time-consuming for (bounded-sub)optimal solvers to solve large MAPF instances. Anytime algorithms find solutions quickly for large instances and then improve them to close-to-optimal ones over time. In this paper, we improve the current state-of-the-art anytime solver MAPF-LNS, that first finds an initial solution fast and then repeatedly replans the paths of subsets of agents via Large Neighborhood Search (LNS). It generates the subsets of agents for replanning by randomized destroy heuristics, but not all of them increase the solution quality substantially. We propose to use machine learning to learn how to select a subset of agents from a collection of subsets, such that replanning increases the solution quality more. We show experimentally that our solver, MAPF-ML-LNS, significantly outperforms MAPF-LNS on the standard MAPF benchmark set in terms of both the speed of improving the solution and the final solution quality.

poster session 4 @ blue 6, poster session 9 @ blue 6, poster session 4, poster session 9

Invariant risk minimization (IRM) has recently emerged as a promising alternative for domain generalization. Nevertheless, the loss function is difficult to optimize for nonlinear classifiers and the original optimization objective could fail when pseudo-invariant features and geometric skews exist. Inspired by IRM, in this paper we propose a novel formulation for domain generalization, dubbed invariant information bottleneck (IIB). IIB aims at minimizing invariant risks for nonlinear classifiers and simultaneously mitigating the impact of pseudo-invariant features and geometric skews. Specifically, we first present a novel formulation for invariant causal prediction via mutual information. Then we adopt the variational formulation of the mutual information to develop a tractable loss function for nonlinear classifiers. To overcome the failure modes of IRM, we propose to minimize the mutual information between the inputs and the corresponding representations. IIB significantly outperforms IRM on synthetic datasets, where the pseudo-invariant features and geometric skews occur, showing the effectiveness of proposed formulation in overcoming failure modes of IRM. Furthermore, experiments on DomainBed show that IIB outperforms $13$ baselines by $0.9\%$ on average across $7$ real datasets.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

Explainable artificial intelligence has rapidly emerged since lawmakers have started requiring interpretable models for safety-critical domains. Concept-based neural networks have arisen as explainable-by-design methods as they leverage human-understandable symbols (i.e. concepts) to predict class memberships. However, most of these approaches focus on the identification of the most relevant concepts but do not provide concise, formal explanations of how such concepts are leveraged by the classifier to make predictions. In this paper, we propose a novel end-to-end differentiable approach enabling the extraction of logic explanations from neural networks using the formalism of First-Order Logic. The method relies on an entropy-based criterion which automatically identifies the most relevant concepts. We consider four different case studies to demonstrate that: (i) this entropy-based criterion enables the distillation of concise logic explanations in safety-critical domains from clinical data to computer vision; (ii) the proposed approach outperforms state-of-the-art white-box models in terms of classification accuracy.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

The deployment of vaccines across the US provides significant defense against serious illness and death from COVID-19. Over 70% of vaccine-eligible Americans are at least partially vaccinated, but there are pockets of the population that are under-vaccinated, such as in rural areas and some demographic groups (e.g. age, race, ethnicity). These pockets are extremely susceptible to the Delta variant, exacerbating the healthcare crisis and increasing the risk of new variants. In this paper, we describe a data-driven model that provides real-time support to Virginia public health officials by recommending mobile vaccination site placement in order to target under-vaccinated populations. Our strategy uses fine-grained mobility data, along with US Census and vaccination uptake data, to identify locations that are most likely to be visited by unvaccinated individuals. We further extend our model to choose locations that maximize vaccine uptake among hesitant groups. We show that the top recommended sites vary substantially across some demographics, demonstrating the value of developing customized recommendation models that integrate fine-grained, heterogeneous data sources. We also validate our recommendations by analyzing the success rates of deployed vaccine sites, and show that sites placed closer to our recommended areas administered higher numbers of doses. Our model is the first of its kind to consider evolving mobility patterns in real-time for suggesting placement strategies customized for different targeted demographic groups.

poster session 4 @ red 6, poster session 9 @ red 6, poster session 4, poster session 9

We present a machine learning system that can quantify fine art paintings with a set of visual elements and principles of art. The formal analysis is fundamental for understanding art, but developing such a system is challenging. Paintings have high visual complexities, but it is also difficult to collect enough training data with direct labels. To resolve these practical limitations, we introduce a novel mechanism, called proxy learning, which learns visual concepts in paintings though their general relation to styles. This framework does not require any visual annotation, but only uses style labels and a general relationship between visual concepts and style. In this paper, we propose a novel proxy model and reformulate four pre-existing methods in the context of proxy learning. Through quantitative and qualitative comparison, we evaluate these methods and compare their effectiveness in quantifying the artistic visual concepts, where the general relationship is estimated by language models; GloVe or BERT. The language modeling is a practical and scalable solution requiring no labeling, but it is inevitably imperfect. We demonstrate how the new proxy model is robust to the imperfection, while the other methods are sensitively affected by it.

poster session 4 @ red 2, poster session 11 @ red 2, poster session 4, poster session 11

Modern artificial intelligence (AI) methods have been used to solve problems that many humans struggle to solve. This opens up new opportunities for knowledge discovery and education. We demonstrate ALLURE, an educational AI system for learning to solve the Rubik’s cube that is designed to help students improve their problem solving skills. ALLURE can both find and explain its own strategies for solving the Rubik’s cube as well as build on user-provided strategies. Collaboration between AI and user happens using visual and natural language modalities.

poster session 4 @ blue 1, poster session 11 @ blue 1, poster session 4, poster session 11

This paper presents TS2Vec, a universal framework for learning representations of time series in an arbitrary semantic level. Unlike existing methods, TS2Vec performs contrastive learning in a hierarchical way over augmented context views, which enables a robust contextual representation for each timestamp. Furthermore, to obtain the representation of an arbitrary sub-sequence in the time series, we can apply a simple aggregation over the representations of corresponding timestamps. We conduct extensive experiments on time series classification tasks to evaluate the quality of time series representations. As a result, TS2Vec achieves significant improvement over existing SOTAs of unsupervised time series representation on 125 UCR datasets and 29 UEA datasets. The learned timestamp-level representations also achieve superior results in time series forecasting and anomaly detection tasks. A linear regression trained on top of the learned representations outperforms previous SOTAs of time series forecasting. Furthermore, we present a simple way to apply the learned representations for unsupervised anomaly detection, which establishes SOTA results in the literature. The source code is publicly available at https://github.com/yuezhihan/ts2vec.

poster session 4 @ red 4, poster session 11 @ red 4, oral session 4 @ red 4, poster session 4, poster session 11, oral session 4

Formulating real-world optimization problems often begins with making predictions from historical data (e.g., an optimizer that aims to recommend fast routes relies upon travel-time predictions). Typically, learning the prediction model used to generate the optimization problem and solving that problem are performed in two separate stages. Recent work has showed how such prediction models can be learned end-to-end by differentiating through the optimization task. Such methods often yield empirical improvements, which are typically attributed to end-to-end making better error tradeoffs than the standard loss function used in a two-stage solution. We refine this explanation and more precisely characterize when end-to-end can improve performance. When prediction targets are stochastic, a two-stage solution must make an a priori choice about which statistics of the target distribution to model---we consider expectations over prediction targets---while an end-to-end solution can make this choice adaptively. We show that the performance gap between a two-stage and end-to-end approach is closely related to the \emph{price of correlation} concept in stochastic optimization and show the implications of some existing POC results for the predict-then-optimize problem. We then consider a novel and particularly practical setting, where multiple prediction targets are combined to obtain each of the objective function’s coefficients. We give explicit constructions where (1) two-stage performs unboundedly worse than end-to-end; and (2) two-stage is optimal. We use simulations to experimentally quantify performance gaps and identify a wide range of real-world applications from the literature whose objective functions rely on multiple prediction targets, suggesting that end-to-end learning could yield significant improvements.

poster session 4 @ blue 4, poster session 8 @ blue 4, poster session 4, poster session 8

Knowledge Graph Question Answering (KGQA) involves retrieving facts from a Knowledge Graph (KG) using natural language queries. A KG is a curated set of facts consisting of entities linked by relations. Certain facts include also temporal information forming a Temporal KG (TKG). Although many natural questions involve explicit or implicit time constraints, question answering (QA) over TKGs has been a relatively unexplored area. Existing solutions are mainly designed for simple temporal questions that can be answered directly by a single TKG fact. This paper puts forth a comprehensive embedding-based framework for answering complex questions over TKGs. Our method termed temporal question reasoning (TempoQR) exploits TKG embeddings to ground the question to the specific entities and time scope it refers to. It does so by augmenting the question embeddings with context, entity and time-aware information by employing three specialized modules. The first computes a textual representation of a given question, the second combines it with the entity embeddings for entities involved in the question, and the third generates question-specific time embeddings. Finally, a transformer-based encoder learns to fuse the generated temporal information with the question representation, which is used for answer predictions. Extensive experiments show that TempoQR improves accuracy by 25--45 percentage points on complex temporal questions over state-of-the-art approaches and it generalizes better to unseen question types.

poster session 4 @ red 5, poster session 11 @ red 5, poster session 4, poster session 11

We present a novel generalized zero-shot algorithm to recognize perceived emotions from gestures. Our task is to map gestures to novel emotion categories not encountered in training. We introduce an adversarial autoencoder-based representation learning that correlates 3D motion-captured gesture sequences with the vectorized representation of the natural-language perceived emotion terms using word2vec embeddings. The language-semantic embedding provides a representation of the emotion label space, and we leverage this underlying distribution to map the gesture sequences to the appropriate categorical emotion labels. We train our method using a combination of gestures annotated with known emotion terms and gestures not annotated with any emotions. We evaluate our method on the MPI Emotional Body Expressions Database (EBEDB) and obtain an accuracy of 58.43%. We see an improvement in performance compared to current state-of-the-art algorithms for generalized zero-shot learning by an absolute 25-27%. We also demonstrate our approach on publicly available online videos and movie scenes, where the actors' pose has been extracted and map to their respective emotive states.

poster session 4 @ red 1, poster session 7 @ red 1, oral session 4 @ red 1, poster session 4, poster session 7, oral session 4

Artificial intelligence (AI) is a promising technology to transform the healthcare industry. Due to the highly sensitive nature of patient data, federated learning (FL) is often leveraged to build models for smart healthcare applications. Existing deployed FL frameworks cannot address the key issues of varying data quality and heterogeneous data distributions across multiple institutions in this sector. In this paper, we report our experience developing and deploying the Contribution-Aware Federated Learning (CAFL) framework for smart healthcare. It provides an efficient and accurate approach to fairly evaluate FL participants' contribution to model performance without exposing their private data, and improves the FL model training protocol to allow the best performing intermediate models to be distributed to participants for FL training. Since its deployment in Yidu Cloud Technology Inc. in March 2021, CAFL has served 8 well-established medical institutions in China to build healthcare decision support models. It can perform contribution evaluations 2.84 times faster than the best existing approach, and has improved the average accuracy of the resulting models by 2.62% compared to the previous system (which is significant in industrial settings). To our knowledge, it is the first contribution-aware federated learning successfully deployed in the healthcare industry.

poster session 4 @ red 6, poster session 9 @ red 6, poster session 4, poster session 9

In online algorithm selection (OAS), instances of an algorithmic problem class are presented to an agent one after another, and the agent has to quickly select a presumably best algorithm from a fixed set of candidate algorithms. For decision problems such as satisfiability (SAT), quality typically refers to the algorithm's runtime. As the latter is known to exhibit a heavy-tail distribution, an algorithm is normally stopped when exceeding a predefined upper time limit. As a consequence, machine learning methods used to optimize an algorithm selection strategy in a data-driven manner need to deal with right-censored samples, a problem that has received little attention in the literature so far. In this work, we revisit multi-armed bandit algorithms for OAS and discuss their capability of dealing with the problem. Moreover, we adapt them towards runtime-oriented losses, allowing for partially censored data while keeping a space- and time-complexity independent of the time horizon. In an extensive experimental evaluation on an adapted version of the ASlib benchmark, we demonstrate that theoretically well-founded methods based on Thompson sampling perform specifically strong and improve in comparison to existing methods.

poster session 4 @ blue 6, poster session 9 @ blue 6, poster session 4, poster session 9

Multi-Agent Path Finding (MAPF), the problem of finding collision-free paths for multiple agents, is important in many applications, such as automated warehouses. EECBS is a leading two-level algorithm that solves MAPF bounded-suboptimally, that is, within some factor $w$ of the optimal cost. \shao{On the low level, EECBS uses focal search to find bounded-suboptimal paths while reducing the number of collisions.} On the high level, EECBS uses explicit estimation search (EES) to resolve collisions. EES keeps track of a lower bound of the optimal cost ($LB$) and finds paths with sum of path costs at most $w \cdot LB$ in order to guarantee bounded suboptimality. However, many paths are typically much shorter than $w$ times their optimal path lengths, meaning that the sum of path costs is much less than $w$ times the optimal cost. In this paper, we propose Flexible EECBS (FEECBS), which uses a flex(ible) distribution of the path cost to relax the bounded-suboptimality restriction on the low level in order to further reduce the number of collisions while still guaranteeing to solve MAPF bounded-suboptimally. Flex distribution can make raising $LB$ harder, so we do not only introduce restrictions for using it but also combine it with A* search to avoid the search being slow. Our empirical evaluation shows that EECBS with flex distribution outperforms on MAPF instances with large maps and large numbers of agents and has competitive performance on MAPF instances with small maps.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7

We consider the problem of acquiring causal representations and concepts in a reinforcement learning setting. Our approach defines a causal variable as being both manipulable by a policy, and able to predict the outcome. We thereby obtain a parsimonious causal graph in which interventions occur at the level of policies. The approach avoids defining a generative model of the data, prior pre-processing, or learning the transition kernel of the Markov decision process. Instead, causal variables and policies are determined by maximizing a new optimization target inspired by mediation analysis, which differs from the expected return. The maximization is accomplished using a generalization of Bellman's equation which is shown to converge, and the method finds meaningful causal representations in a simulated environment.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

Event logs are often one of the main sources of information to understand the behavior of a system. While numerous approaches have extracted partial information from event logs, in this work, we aim at inferring a global model of a system from its event logs. We consider real-time systems, which can be modeled with Timed Automata: our approach is thus a Timed Automata learner. There is a handful of related work, however, they might require a lot of parameters or produce Timed Automata that either are undeterministic or lack precision. In contrast, our proposed approach, called TAG, requires only one parameter and learns a deterministic Timed Automaton having a good tradeoff between accuracy and complexity of the automata. This allows getting an interpretable and accurate global model of the real-time system considered. Our experiments compare our approach to the related work and demonstrate its merits.

poster session 4 @ blue 2, poster session 7 @ blue 2, poster session 4, poster session 7

Existing image semantic segmentation methods favor learning consistent representations by extracting long-range contextual features with the attention, multi-scale, or graph aggregation strategies. These methods usually treat the misclassified and correctly classified pixels equally, hence misleading the optimization process and causing inconsistent intra-class pixel feature representations in the embedding space during learning. In this paper, we propose the auxiliary representation calibration head (RCH), which consists of the image decoupling, prototype clustering, error calibration modules and a metric loss function, to calibrate these error-prone feature representations for better intra-class consistency and segmentation performance. RCH could be incorporated into the hidden layers, trained together with the segmentation networks, and decoupled in the inference stage without additional parameters. Experimental results show that our method could significantly boost the performance of current segmentation methods on multiple datasets (e.g., we outperform the original HRNet and OCRNet by 1.1% and 0.9% mIoU on the Cityscapes test set). Codes are available at https://github.com/VipaiLab/RCH.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

Optimization is often cast as a deterministic problem, where the solution is found through some iterative procedure such as gradient descent. However, when training neural networks the loss function changes over (iteration) time due to the randomized selection of a subset of the samples. This randomization turns the optimization problem into a stochastic one. We propose to consider the loss as a noisy observation with respect to some reference optimum. This interpretation of the loss allows us to adopt Kalman filtering as an optimizer, as its recursive formulation is designed to estimate unknown parameters from noisy measurements. Moreover, we show that the Kalman Filter dynamical model for the evolution of the unknown parameters can be used to capture the gradient dynamics of advanced methods such as Momentum and Adam. We call this stochastic optimization method KOALA, which is short for Kalman Optimization Algorithm with Loss Adaptivity. KOALA is an easy to implement, scalable, and efficient method to train neural networks. We provide convergence analysis and show experimentally that it yields parameter estimates that are on par with or better than existing state of the art optimization algorithms across several neural network architectures and machine learning tasks, such as computer vision and language modeling.

poster session 4 @ blue 3, poster session 9 @ blue 3, poster session 4, poster session 9

Mean Field Games (MFGs) can potentially scale multi-agent systems to extremely large populations of agents. Yet, most of the literature assumes a single initial distribution for the agents, which limits the practical applications of MFGs. Ma- chine Learning has the potential to solve a wider diversity of MFG problems thanks to generalizations capacities. We study how to leverage these generalization properties to learn policies enabling a typical agent to behave optimally against any population distribution. In reference to the Master equation in MFGs, we coin the term “Master policies” to describe them and we prove that a single Master policy provides a Nash equilibrium, whatever the initial distribution. We propose a method to learn such Master policies. Our approach relies on three ingredients: adding the current population distribution as part of the observation, approximating Master policies with neural networks, and training via Reinforcement Learning and Fictitious Play. We illustrate on numerical examples not only the efficiency of the learned Master policy but also its generalization capabilities beyond the distributions used for training.

poster session 4 @ red 3, poster session 8 @ red 3, poster session 4, poster session 8

Image operators have been extensively applied to create visually attractive photos for users to share processed images on social media. However, most image operators often smooth out details or generate textures after the processing, which removes the original content and raises challenges for restoring the original image. To resolve this issue, we propose a quasi-invertible model that learns common image processing operators in a restorable fashion: the learned image operators can generate visually pleasing results with the original content embedded. Our model is trained on input-output pairs that represent an image processing operator's behavior and uses a network that consists of an invertible branch and a non-invertible branch to increase our model's approximation capability. We evaluate the proposed model on ten image operators, including detail enhancement, abstraction, blur, photographic style, and non-photorealistic style. Extensive experiments show that our approach outperforms relevant baselines in the restoration quality, and the learned restorable operator is fast in inference and robust to compression. Furthermore, we demonstrate that the invertible operator can be easily applied to practical applications such as restorable human face retouching and highlight preserved exposure adjustment.

poster session 4 @ red 1, poster session 7 @ red 1, poster session 4, poster session 7