IJCAI 2020 论文列表

Students’ gestures, hand-raising, stand-up, and sleeping, indicates the engagement of students in classrooms and partially reflects teaching quality. Therefore, fast and automatically recognizing these gestures are of great importance. Due to limited computational resources in primary and secondary schools, we propose a real-time student behavior detector based on light-weight MobileNetV2-SSD to reduce the dependency of GPUs. Firstly, we build a large-scale corpus from real schools to capture various behavior gestures. Based on such a corpus, we transfer the gesture recognition task into object detections. Secondly, we design a multi-dimensional attention-based detector, named GestureDet, for real-time and accurate gesture analysis. The multi-dimensional attention mechanisms simultaneously consider all the dimensions of the training set, aiming to pay more attention to discriminative features and samples that are important for the final performance. Specifically, the spatial attention is constructed with stacked dilated convolution layers to generate a soft and learnable mask for re-weighting foreground and background features; the channel attention introduces the context modeling and squeeze-and-excitation module to focus on discriminative features; the batch attention discriminates important samples with a new designed reweight strategy. Experimental results demonstrate the effectiveness and versatility of GestureDet, which achieves 75.2% mAP on real student behavior dataset, and 74.5% on public PASCAL VOC dataset at 20fps on embedding device Nvidia Jetson TX2. Code will be made publicly available.

Audio adversarial examples, imperceptible to humans, have been constructed to attack automatic speech recognition (ASR) systems. However, the adversarial examples generated by existing approaches usually incorporate noticeable noises, especially during the periods of silences and pauses. Moreover, the added noises often break temporal dependency property of the original audio, which can be easily detected by state-of-the-art defense mechanisms. In this paper, we propose a new Iterative Proportional Clipping (IPC) algorithm that preserves temporal dependency in audios for generating more robust adversarial examples. We are motivated by an observation that the temporal dependency in audios imposes a significant effect on human perception. Following our observation, we leverage a proportional clipping strategy to reduce noise during the low-intensity periods. Experimental results and user study both suggest that the generated adversarial examples can significantly reduce human-perceptible noises and resist the defenses based on the temporal structure.

Due to the improvement of Language Modeling, the emerging NLP assistant tools aiming for text generation greatly reduce the human workload on writing documents. However, the generation of legal text faces greater challenges than ordinary texts because of its high requirement for keeping logic reasonable, which can not be guaranteed by Language Modeling right now. To generate reasonable legal documents, we propose a novel method CoLMQA, which (1) combines Language Modeling and Question Answering, (2) generates text with slots by Language Modeling, and (3) fills the slots by our proposed Question Answering method named Transformer-based Key-Value Memory Networks. In CoLMQA, the slots represent the text part that needs to be highly constrained by logic, such as the name of the law and the number of the law article. And the Question Answering fills the slots in context with the help of Legal Knowledge Base to keep logic reasonable. The experiment verifies the quality of legal documents generated by CoLMQA, surpassing the documents generated by pure Language Modeling.

Generation of high-quality person images is challenging, due to the sophisticated entanglements among image factors, e.g., appearance, pose, foreground, background, local details, global structures, etc. In this paper, we present a novel end-to-end framework to generate realistic person images based on given person poses and appearances. The core of our framework is a novel generator called Appearance-aware Pose Stylizer (APS) which generates human images by coupling the target pose with the conditioned person appearance progressively. The framework is highly flexible and controllable by effectively decoupling various complex person image factors in the encoding phase, followed by re-coupling them in the decoding phase. In addition, we present a new normalization method named adaptive patch normalization, which enables region-specific normalization and shows a good performance when adopted in person image generation model. Experiments on two benchmark datasets show that our method is capable of generating visually appealing and realistic-looking results using arbitrary image and pose inputs.

Counterfactual explanations help users to understand the behaviors of machine learning models by changing the inputs for the existing outputs. For an image classification task, an example counterfactual visual explanation explains: "for an example that belongs to class A, what changes do we need to make to the input so that the output is more inclined to class B." Our research considers changing the attribute description text of class A on the basis of the attributes of class B and generating counterfactual images on the basis of the modified text. We can use the prediction results of the model on counterfactual images to find the attributes that have the greatest effect when the model is predicting classes A and B. We applied our method to a fine-grained image classification dataset and used the generative adversarial network to generate natural counterfactual visual explanations. To evaluate these explanations, we used them to assist crowdsourcing workers in an image classification task. We found that, within a specific range, they improved classification accuracy.

Accurate local-level poverty measurement is an essential task for governments and humanitarian organizations to track the progress towards improving livelihoods and distribute scarce resources. Recent computer vision advances in using satellite imagery to predict poverty have shown increasing accuracy, but they do not generate features that are interpretable to policymakers, inhibiting adoption by practitioners. Here we demonstrate an interpretable computational framework to accurately predict poverty at a local level by applying object detectors to high resolution (30cm) satellite images. Using the weighted counts of objects as features, we achieve 0.539 Pearson's r^2 in predicting village-level poverty in Uganda, a 31% improvement over existing (and less interpretable) benchmarks. Feature importance and ablation analysis reveal intuitive relationships between object counts and poverty predictions. Our results suggest that interpretability does not have to come at the cost of performance, at least in this important domain.

Generating diverse behaviors for game artificial intelligence (Game AI) has been long recognized as a challenging task in the game industry. Designing a Game AI with a satisfying behavioral characteristic (style) heavily depends on the domain knowledge and is hard to achieve manually. Deep reinforcement learning sheds light on advancing the automatic Game AI design. However, most of them focus on creating a superhuman Game AI, ignoring the importance of behavioral diversity in games. To bridge the gap, we introduce a new framework, named EMOGI, which can automatically generate desirable styles with almost no domain knowledge. More importantly, EMOGI succeeds in creating a range of diverse styles, providing behavior-diverse Game AIs. Evaluations on the Atari and real commercial games indicate that, compared to existing algorithms, EMOGI performs better in generating diverse behaviors and significantly improves the efficiency of Game AI design.

The International Classification of Diseases (ICD) is a list of classification codes for the diagnoses. Automatic ICD coding is a multi-label text classification problem with noisy clinical document inputs and long-tailed label distribution, making it difficult for fine-grained classification on both frequent and zero-shot codes at the same time, i.e. generalized zero-shot ICD coding. In this paper, we propose a latent feature generation framework to improve the prediction on unseen codes without compromising the performance on seen codes. Our framework generates semantically meaningful features for zero-shot codes by exploiting ICD code hierarchical structure and reconstructing the code-relevant keywords with a novel cycle architecture. To the best of our knowledge, this is the first adversarial generative model for generalized zero-shot learning on multi-label text classification. Extensive experiments demonstrate the effectiveness of our approach. On the public MIMIC-III dataset, our methods improve the F1 score from nearly 0 to 20.91% for the zero-shot codes, and increase the AUC score by 3% (absolute improvement) from previous state of the art. Code is available at https://github.com/csong27/gzsl_text.

Deep reinforcement learning (DRL) increases the successful applications of reinforcement learning (RL) techniques but also brings challenges such as low sample efficiency. In this work, I propose generalized representation learning methods to obtain compact state space suitable for RL from a raw observation state. I expect my new methods will increase sample efficiency of RL by understandable representations of state and therefore improve the performance of RL.

We consider Monte-Carlo Tree Search (MCTS) applied to Markov Decision Processes (MDPs) and Partially Observable MDPs (POMDPs), and the well-known Upper Confidence bound for Trees (UCT) algorithm. In UCT, a tree with nodes (states) and edges (actions) is incrementally built by the expansion of nodes, and the values of nodes are updated through a backup strategy based on the average value of child nodes. However, it has been shown that with enough samples the maximum operator yields more accurate node value estimates than averaging. Instead of settling for one of these value estimates, we go a step further proposing a novel backup strategy which uses the power mean operator, which computes a value between the average and maximum value. We call our new approach Power-UCT, and argue how the use of the power mean operator helps to speed up the learning in MCTS. We theoretically analyze our method providing guarantees of convergence to the optimum. Finally, we empirically demonstrate the effectiveness of our method in well-known MDP and POMDP benchmarks, showing significant improvement in performance and convergence speed w.r.t. state of the art algorithms.

Maximum likelihood learning is a well-studied approach for fitting discrete Markov random fields (MRFs) to data. However, general purpose maximum likelihood estimation for fitting MRFs with continuous variables have only been studied in much more limited settings. In this work, we propose a generic MLE estimation procedure for MRFs whose potential functions are modeled by neural networks. To make learning effective in practice, we show how to leverage a highly parallelizable variational inference method that can easily fit into popular machining learning frameworks like TensorFlow. We demonstrate experimentally that our approach is capable of effectively modeling the data distributions of a variety of real data sets and that it can compete effectively with other common methods on multilabel classification and generative modeling tasks.

The process of belief revision occurs in many applications where agents may have incorrect or incomplete information. One important theoretical model of belief revision is the well-known AGM approach. Unfortunately, there are few tools available for solving AGM revision problems quickly; this has limited the use of AGM operators for practical applications. In this demonstration paper, we describe GenC, a tool that is able to quickly calculate the result of AGM belief revision for formulas with hundreds of variables and millions of clauses. GenC uses an AllSAT solver and parallel processing to solve revision problems at a rate much faster than existing systems. The solver works for the class of parametrised difference operators, which is an extensive class of revision operators that use a weighted Hamming distance to measure the similarity between states. We demonstrate how GenC can be used as a stand-alone tool or as a component of a reasoning system for a variety of applications.

Gaussian Embedding of Linked Documents (GELD) is a new method that embeds linked documents (e.g., citation networks) onto a pretrained semantic space (e.g., a set of word embeddings). We formulate the problem in such a way that we model each document as a Gaussian distribution in the word vector space. We design a generative model that combines both words and links in a consistent way. Leveraging the variance of a document allows us to model the uncertainty related to word and link generation. In most cases, our method outperforms state-of-the-art methods when using our document vectors as features for usual downstream tasks. In particular, GELD achieves better accuracy in classification and link prediction on Cora and Dblp. In addition, we demonstrate qualitatively the convenience of several properties of our method. We provide the implementation of GELD and the evaluation datasets to the community (https://github.com/AntoineGourru/DNEmbedding).

Authorship verification is an important problem that has many applications. The state-of-the-art deep authorship verification methods typically leverage character-level language models to encode author-specific writing styles. However, they often fail to capture syntactic level patterns, leading to sub-optimal accuracy in cross-topic scenarios. Also, due to imperfect cross-author parameter sharing, it's difficult for them to distinguish author-specific writing style from common patterns, leading to data-inefficient learning. This paper introduces a novel POS-level (Part of Speech) gated RNN based language model to effectively learn the author-specific syntactic styles. The author-agnostic syntactic information obtained from the POS tagger pre-trained on large external datasets greatly reduces the number of effective parameters of our model, enabling the model to learn accurate author-specific syntactic styles with limited training data. We also utilize a gated architecture to learn the common syntactic writing styles with a small set of shared parameters and let the author-specific parameters focus on each author's special syntactic styles. Extensive experimental results show that our method achieves significantly better accuracy than state-of-the-art competing methods, especially in cross-topic scenarios (over 5\% in terms of AUC-ROC).

The popular tracking-by-detection paradigm for multi-object tracking (MOT) focuses on solving data association problem, of which a robust similarity model lies in the heart. Most previous works make effort to improve feature representation for individual object while leaving the relations among objects less explored, which may be problematic in some complex scenarios. In this paper, we focus on leveraging the relations among objects to improve robustness of the similarity model. To this end, we propose a novel graph representation that takes both the feature of individual object and the relations among objects into consideration. Besides, a graph matching module is specially designed for the proposed graph representation to alleviate the impact of unreliable relations. With the help of the graph representation and the graph matching module, the proposed graph similarity model, named GSM, is more robust to the occlusion and the targets sharing similar appearance. We conduct extensive experiments on challenging MOT benchmarks and the experimental results demonstrate the effectiveness of the proposed method.

Facial action unit (AU) intensity estimation aims to measure the intensity of different facial muscle movements. The external knowledge such as AU co-occurrence relationship is typically leveraged to improve performance. However, the AU characteristics may vary among individuals due to different physiological structures of human faces. To this end, we propose a novel geometry-guided representation learning (G2RL) method for facial AU intensity estimation. Specifically, our backbone model is based on a heatmap regression framework, where the produced heatmaps reflect rich information associated with AU intensities and their spatial distributions. Besides, we incorporate the external geometric knowledge into the backbone model to guide the training process via a learned projection matrix. The experimental results on two benchmark datasets demonstrate that our method is comparable with the state-of-the-art approaches, and validate the effectiveness of incorporating external geometric knowledge for facial AU intensity estimation.

Neural network compression and quantization are important tasks for fitting state-of-the-art models into the computational, memory and power constraints of mobile devices and embedded hardware. Recent approaches to model compression/quantization are based on reinforcement learning or search methods to quantize the neural network for a specific hardware platform. However, these methods require multiple runs to compress/quantize the same base neural network to different hardware setups. In this work, we propose a fully nested neural network (FN3) that runs only once to build a nested set of compressed/quantized models, which is optimal for different resource constraints. Specifically, we exploit the additive characteristic in different levels of building blocks in neural network and propose an ordered dropout (ODO) operation that ranks the building blocks. Given a trained FN3, a fast heuristic search algorithm is run offline to find the optimal removal of components to maximize the accuracy under different constraints. Compared with the related works on adaptive neural network designed only for channels or bits, the proposed approach is applicable to different levels of building blocks (bits, neurons, channels, residual paths and layers). Empirical results validate strong practical performance of proposed approach.

Although symbolic bidirectional search is successful in optimal classical planning, state-of-the-art satisficing planners do not use bidirectional search. Previous bidirectional search planners for satisficing planning behaved similarly to a trivial portfolio, which independently executes forward and backward search without the desired ``meet-in-the-middle'' behavior of bidirectional search where the forward and backward search frontiers intersect at some point relatively far from the forward and backward start states. In this paper, we propose Top-to-Top Bidirectional Search (TTBS), a new bidirectional search strategy with front-to-front heuristic evaluation. We show that TTBS strongly exhibits ``meet-in-the-middle'' behavior and can solve instances solved by neither forward nor backward search on a number of domains.

The distinctive driving force of constraint programming (CP) to solve combinatorial problems has been a privileged access to problem structure through the high-level models it uses. We investigate a richer propagation medium for CP made possible by recent work on counting solutions inside constraints. Beliefs about individual variable-value assignments are exchanged between contraints and iteratively adjusted. Its advantage over standard belief propagation is that the higher-level models do not tend to create as many cycles, which are known to be problematic for convergence. We find that it significantly improves search guidance.