IJCAI 2020 Papers

Document-level label consistency is an effective indicator that different occurrences of a particular token sequence are very likely to have the same entity types. Previous work focused on better context representations and used the CRF for label decoding. However, CRF-based methods are inadequate for modeling document-level label consistency. This work introduces a novel two-stage label refinement approach to handle document-level label consistency, where a key-value memory network is first used to record draft labels predicted by the base model, and then a multi-channel Transformer makes refinements on these draft predictions based on the explicit co-occurrence relationship derived from the memory network. In addition, in order to mitigate the side effects of incorrect draft labels, Bayesian neural networks are used to indicate the labels with a high probability of being wrong, which can greatly assist in preventing the incorrect refinement of correct draft labels. The experimental results on three named entity recognition benchmarks demonstrated that the proposed method significantly outperformed the state-of-the-art methods.

Fine-grained entity typing (FET) is a fundamental task for various entity-leveraging applications. Although great success has been made, existing systems still have challenges in handling noisy samples in training data introduced by distant supervision methods. To address these noise, previous studies either focus on processing the clean samples (i,e., have only one label) and noisy samples (i,e., have multiple labels) with different strategies or filtering the noisy labels based on the assumption that the distantly-supervised label set certainly contains the correct type label. In this paper, we propose a probabilistic automatic relabeling method which treats all training samples uniformly. Our method aims to estimate the pseudo-truth label distribution of each sample, and the pseudo-truth distribution will be treated as part of trainable parameters which are jointly updated during the training process. The proposed approach does not rely on any prerequisite or extra supervision, making it effective on real applications. Experiments on several benchmarks show that our method outperforms previous approaches and alleviates the noisy labeling problem.

Due to the difficulty of comprehensive data collection, created by factors such as privacy protection and sensor device limitations, we often need to analyze incomplete transition data where some information is missing from the ideal (complete) transition data. In this paper, we propose a new method that can estimate, in a unified manner, Markov chain parameters from incomplete transition data that consist of hidden transition data (data from which visited state information is partially hidden) and dropped transition data (data from which some state visits are dropped). A key to developing the method is regarding the hidden and dropped transition data as labeled and unlabeled multi-step transition data, where the labels represent the number of steps required for each transition. This allows us to describe the generative process of multi-step transition data, and thus develop a new probabilistic model. We confirm the effectiveness of the proposal by experiments on synthetic and real data.

Generating natural language descriptions for videos, i.e., video captioning, essentially requires step-by-step reasoning along the generation process. For example, to generate the sentence “a man is shooting a basketball”, we need to first locate and describe the subject “man”, next reason out the man is “shooting”, then describe the object “basketball” of shooting. However, existing visual reasoning methods designed for visual question answering are not appropriate to video captioning, for it requires more complex visual reasoning on videos over both space and time, and dynamic module composition along the generation process. In this paper, we propose a novel visual reasoning approach for video captioning, named Reasoning Module Networks (RMN), to equip the existing encoder-decoder framework with the above reasoning capacity. Specifically, our RMN employs 1) three sophisticated spatio-temporal reasoning modules, and 2) a dynamic and discrete module selector trained by a linguistic loss with a Gumbel approximation. Extensive experiments on MSVD and MSR-VTT datasets demonstrate the proposed RMN outperforms the state-of-the-art methods while providing an explicit and explainable generation process. Our code is available at https://github.com/tgc1997/RMN.

A rising vision for AI in the open world centers on the development of systems that can complement humans for perceptual, diagnostic, and reasoning tasks. To date, systems aimed at complementing the skills of people have employed models trained to be as accurate as possible in isolation. We demonstrate how an end-to-end learning strategy can be harnessed to optimize the combined performance of human-machine teams by considering the distinct abilities of people and machines. The goal is to focus machine learning on problem instances that are difficult for humans, while recognizing instances that are difficult for the machine and seeking human input on them. We demonstrate in two real-world domains (scientific discovery and medical diagnosis) that human-machine teams built via these methods outperform the individual performance of machines and people. We then analyze conditions under which this complementarity is strongest, and which training methods amplify it. Taken together, our work provides the first systematic investigation of how machine learning systems can be trained to complement human reasoning.

Bipartite b-matching is fundamental in algorithm design, and has been widely applied into diverse applications, such as economic markets, labor markets, etc. These practical problems usually exhibit two distinct features: large-scale and dynamic, which requires the matching algorithm to be repeatedly executed at regular intervals. However, existing exact and approximate algorithms usually fail in such settings due to either requiring intolerable running time or too much computation resource. To address this issue, based on a key observation that the matching instances vary not too much, we propose NeuSearcher which leverage the knowledge learned from previously instances to solve new problem instances. Specifically, we design a multichannel graph neural network to predict the threshold of the matched edges, by which the search region could be significantly reduced. We further propose a parallel heuristic search algorithm to iteratively improve the solution quality until convergence. Experiments on both open and industrial datasets demonstrate that NeuSearcher can speed up 2 to 3 times while achieving exactly the same matching solution compared with the state-of-the-art approximation approaches.

Complementary recommendations, which aim at providing users product suggestions that are supplementary and compatible with their obtained items, have become a hot topic in both academia and industry in recent years. Existing work mainly focused on modeling the co-purchased relations between two items, but the compositional associations of item collections are largely unexplored. Actually, when a user chooses the complementary items for the purchased products, it is intuitive that she will consider the visual semantic coherence (such as color collocations, texture compatibilities) in addition to global impressions. Towards this end, in this paper, we propose a novel Content Attentive Neural Network (CANN) to model the comprehensive compositional coherence on both global contents and semantic contents. Specifically, we first propose a Global Coherence Learning (GCL) module based on multi-heads attention to model the global compositional coherence. Then, we generate the semantic-focal representations from different semantic regions and design a Focal Coherence Learning (FCL) module to learn the focal compositional coherence from different semantic-focal representations. Finally, we optimize the CANN in a novel compositional optimization strategy. Extensive experiments on the large-scale real-world data clearly demonstrate the effectiveness of CANN compared with several state-of-the-art methods.

The ability to learn from human supervision is fundamental for personal assistants and other interactive applications of AI. Two central challenges for deploying interactive learners in the wild are the unreliable nature of the supervision and the varying complexity of the prediction task. We address a simple but representative setting, incremental classification in the wild, where the supervision is noisy and the number of classes grows over time. In order to tackle this task, we propose a redesign of skeptical learning centered around Gaussian Processes (GPs). Skeptical learning is a recent interactive strategy in which, if the machine is sufficiently confident that an example is mislabeled, it asks the annotator to reconsider her feedback. In many cases, this is often enough to obtain clean supervision. Our redesign, dubbed ISGP , leverages the uncertainty estimates supplied by GPs to better allocate labeling and contradiction queries, especially in the presence of noise. Our experiments on synthetic and real-world data show that, as a result, while the original formulation of skeptical learning produces over-confident models that can fail completely in the wild, ISGP works well at varying levels of noise and as new classes are observed.

Despite the huge progress in scene graph generation in recent years, its long-tail distribution in object relationships remains a challenging and pestering issue. Existing methods largely rely on either external knowledge or statistical bias information to alleviate this problem. In this paper, we tackle this issue from another two aspects: (1) scene-object interaction aiming at learning specific knowledge from a scene via an additive attention mechanism; and (2) long-tail knowledge transfer which tries to transfer the rich knowledge learned from the head into the tail. Extensive experiments on the benchmark dataset Visual Genome on three tasks demonstrate that our method outperforms current state-of-the-art competitors. Our source code is available at https://github.com/htlsn/issg.

Learning from imbalanced data, where the positive examples are very scarce, remains a challenging task from both a theoretical and algorithmic perspective. In this paper, we address this problem using a metric learning strategy. Unlike the state-of-the-art methods, our algorithm MLFP, for Metric Learning from Few Positives, learns a new representation that is used only when a test query is compared to a minority training example. From a geometric perspective, it artificially brings positive examples closer to the query without changing the distances to the negative (majority class) data. This strategy allows us to expand the decision boundaries around the positives, yielding a better F-Measure, a criterion which is suited to deal with imbalanced scenarios. Beyond the algorithmic contribution provided by MLFP, our paper presents generalization guarantees on the false positive and false negative rates. Extensive experiments conducted on several imbalanced datasets show the effectiveness of our method.

This survey gives a selective review of recent development of machine learning (ML) for combinatorial optimization (CO), especially for graph matching. The synergy of these two well-developed areas (ML and CO) can potentially give transformative change to artificial intelligence, whose foundation relates to these two building blocks. For its representativeness and wide-applicability, this paper is more focused on the problem of weighted graph matching, especially from the learning perspective. For graph matching, we show that many learning techniques e.g. convolutional neural networks, graph neural networks, reinforcement learning can be effectively incorporated in the paradigm for extracting the node features, graph structure features, and even the matching engine. We further present outlook for the new settings for learning graph matching, and direction towards more integrated combinatorial optimization solvers with prediction models, and also the mutual embrace of traditional solver and machine learning components.

Regular Decision Processes (RDPs) are a recently introduced model that extends MDPs with non-Markovian dynamics and rewards. The non-Markovian behavior is restricted to depend on regular properties of the history. These can be specified using regular expressions or formulas in linear dynamic logic over finite traces. Fully specified RDPs can be solved by compiling them into an appropriate MDP. Learning RDPs from data is a challenging problem that has yet to be addressed, on which we focus in this paper. Our approach rests on a new representation for RDPs using Mealy Machines that emit a distribution and an expected reward for each state-action pair. Building on this representation, we combine automata learning techniques with history clustering to learn such a Mealy machine and solve it by adapting MCTS to it. We empirically evaluate this approach, demonstrating its feasibility.

Subquadratic norms have been studied recently in the context of structured sparsity, which has been shown to be more beneficial than conventional regularizers in applications such as image denoising, compressed sensing, banded covariance estimation, etc. While existing works have been successful in learning structured sparse models such as trees, graphs, their associated optimization procedures have been inefficient because of hard-to-evaluate proximal operators of the norms. In this paper, we study the computational aspects of learning with subquadratic norms in a general setup. Our main contributions are two proximal-operator based algorithms ADMM-η and CP-η, which generically apply to these learning problems with convex loss functions, and achieve a proven rate of convergence of O(1/T) after T iterations. These algorithms are derived in a primal-dual framework, which have not been examined for subquadratic norms. We illustrate the efficiency of the algorithms developed in the context of tree-structured sparsity, where they comprehensively outperform relevant baselines.

A Linked Data crawler performs a selection to focus on collecting linked RDF (including RDFa) data on the Web. From the perspectives of throughput and coverage, given a newly discovered and targeted URI, the key issue of Linked Data crawlers is to decide whether this URI is likely to dereference into an RDF data source and therefore it is worth downloading the representation it points to. Current solutions adopt heuristic rules to filter irrelevant URIs. But when the heuristics are too restrictive this hampers the coverage of crawling. In this paper, we propose and compare approaches to learn strategies for crawling Linked Data on the Web by predicting whether a newly discovered URI will lead to an RDF data source or not. We detail the features used in predicting the relevance and the methods we evaluated including a promising adaptation of FTRL-proximal online learning algorithm. We compare several options through extensive experiments including existing crawlers as baseline methods to evaluate their efficiency.

Few-shot learning for visual recognition aims to adapt to novel unseen classes with only a few images. Recent work, especially the work based on low-level information, has achieved great progress. In these work, local representations (LRs) are typically employed, because LRs are more consistent among the seen and unseen classes. However, most of them are limited to an individual image-to-image or image-to-class measure manner, which cannot fully exploit the capabilities of LRs, especially in the context of a certain task. This paper proposes an Adaptive Task-aware Local Representations Network (ATL-Net) to address this limitation by introducing episodic attention, which can adaptively select the important local patches among the entire task, as the process of human recognition. We achieve much superior results on multiple benchmarks. On the miniImagenet, ATL-Net gains 0.93% and 0.88% improvements over the compared methods under the 5-way 1-shot and 5-shot settings. Moreover, ATL-Net can naturally tackle the problem that how to adaptively identify and weight the importance of different key local parts, which is the major concern of fine-grained recognition. Specifically, on the fine-grained dataset Stanford Dogs, ATL-Net outperforms the second best method with 5.39% and 9.69% gains under the 5-way 1-shot and 5-shot settings.

Deep neural networks perform well on test data when they are highly overparameterized, which, however, also leads to large cost to train and deploy them. As a leading approach to address this problem, sparse neural networks have been widely used to significantly reduce the size of networks, making them more efficient during training and deployment, without compromising performance. Recently, sparse neural networks, either compressed from a pre-trained model or obtained by training from scratch, have been observed to be able to generalize as well as or even better than their dense counterparts. However, conventional techniques to find well fitted sparse sub-networks are expensive and the mechanisms underlying this phenomenon are far from clear. To tackle these problems, this Ph.D. research aims to study the generalization of sparse neural networks, and to propose more efficient approaches that can yield sparse neural networks with generalization bounds.

Solving the problem is an important part of optimization. An equally important part is the analysis of the solution where several questions can arise. For a scheduling problem, is it possible to obtain a better solution by increasing the capacity of a resource? What happens to the objective value if we start a specific task earlier? Answering such questions is important to provide explanations and increase the acceptability of a solution. A lot of research has been done on sensitivity analysis, but few techniques can be applied to constraint programming. We present a new method for sensitivity analysis applied to constraint programming. It collects information, during the search, about the propagation of the CUMULATIVE constraint, the filtering of the variables, and the solution returned by the solver. Using machine learning algorithms, we predict if increasing/decreasing the capacity of the cumulative resource allows a better solution. We also predict the impact on the objective value of forcing a task to finish earlier. We experimentally validate our method with the RCPSP problem.

Recent deep learning-based Brain-Computer Interface (BCI) decoding algorithms mainly focus on spatial-temporal features, while failing to explicitly explore spectral information which is one of the most important cues for BCI. In this paper, we propose a novel regional attention convolutional neural network (RACNN) to take full advantage of spectral-spatial-temporal features for EEG motion intention recognition. Time-frequency based analysis is adopted to reveal spectral-temporal features in terms of neural oscillations of primary sensorimotor. The basic idea of RACNN is to identify the activated area of the primary sensorimotor adaptively. The RACNN aggregates a varied number of spectral-temporal features produced by a backbone convolutional neural network into a compact fixed-length representation. Inspired by the neuroscience findings that functional asymmetry of the cerebral hemisphere, we propose a region biased loss to encourage high attention weights for the most critical regions. Extensive evaluations on two benchmark datasets and real-world BCI dataset show that our approach significantly outperforms previous methods.

Cross-domain recommendation methods usually transfer knowledge across different domains implicitly, by sharing model parameters or learning parameter mappings in the latent space. Differing from previous studies, this paper focuses on learning explicit mapping between a user's behaviors (i.e. interaction itemsets) in different domains during the same temporal period. In this paper, we propose a novel deep cross-domain recommendation model, called Cycle Generation Networks (CGN). Specifically, CGN employs two generators to construct the dual-direction personalized itemset mapping between a user's behaviors in two different domains over time. The generators are learned by optimizing the distance between the generated itemset and the real interacted itemset, as well as the cycle-consistent loss defined based on the dual-direction generation procedure. We have performed extensive experiments on real datasets to demonstrate the effectiveness of the proposed model, comparing with existing single-domain and cross-domain recommendation methods.