IJCAI 2021 论文列表

Many decision-making processes involve solving a combinatorial optimization problem with uncertain input that can be estimated from historic data. Recently, problems in this class have been successfully addressed via end-to-end learning approaches, which rely on solving one optimization problem for each training instance at every epoch. In this context, we provide two distinct contributions. First, we use a Noise Contrastive approach to motivate a family of surrogate loss functions, based on viewing non-optimal solutions as negative examples. Second, we address a major bottleneck of all predict-and-optimize approaches, i.e. the need to frequently recompute optimal solutions at training time. This is done via a solver-agnostic solution caching scheme, and by replacing optimization calls with a lookup in the solution cache. The method is formally based on an inner approximation of the feasible space and, combined with a cache lookup strategy, provides a controllable trade-off between training time and accuracy of the loss approximation. We empirically show that even a very slow growth rate is enough to match the quality of state-of-the-art methods, at a fraction of the computational cost.

Deep neural networks have achieved outstanding performance in many machine learning tasks. However, this remarkable success is achieved in a closed and static environment where the model is trained using large training data of a single task and deployed for testing on data with a similar distribution. Once the model is deployed, it becomes fixed and inflexible to new knowledge. This contradicts real-world applications, in which agents interact with open and dynamic environments and deal with non-stationary data. This Ph.D. research aims to propose efficient approaches that can develop intelligent agents capable of accumulating new knowledge and adapting to new environments without forgetting the previously learned ones.

Camouflaged object detection (COD) is a challenging task due to the low boundary contrast between the object and its surroundings. In addition, the appearance of camouflaged objects varies significantly, e.g., object size and shape, aggravating the difficulties of accurate COD. In this paper, we propose a novel Context-aware Crosslevel Fusion Network (C2F-Net) to address the challenging COD task. Specifically, we propose an Attention-induced Cross-level Fusion Module (ACFM) to integrate the multi-level features with informative attention coefficients. The fused features are then fed to the proposed Dual-branch Global Context Module (DGCM), which yields multi-scale feature representations for exploiting rich global context information. In C2F-Net, the two modules are conducted on high-level features using a cascaded manner. Extensive experiments on three widely used benchmark datasets demonstrate that our C2F-Net is an effective COD model and outperforms state-of-the-art models remarkably. Our code is publicly available at: https://github.com/thograce/C2FNet.

Recent advances in image inpainting have shown impressive results for generating plausible visual details on rather simple backgrounds. However, for complex scenes, it is still challenging to restore reasonable contents as the contextual information within the missing regions tends to be ambiguous. To tackle this problem, we introduce pretext tasks that are semantically meaningful to estimating the missing contents. In particular, we perform knowledge distillation on pretext models and adapt the features to image inpainting. The learned semantic priors ought to be partially invariant between the high-level pretext task and low-level image inpainting, which not only help to understand the global context but also provide structural guidance for the restoration of local textures. Based on the semantic priors, we further propose a context-aware image inpainting model, which adaptively integrates global semantics and local features in a unified image generator. The semantic learner and the image generator are trained in an end-to-end manner. We name the model SPL to highlight its ability to learn and leverage semantic priors. It achieves the state of the art on Places2, CelebA, and Paris StreetView datasets

Relation Extraction is key to many downstream tasks. Dialogue relation extraction aims at discovering entity relations from multi-turn dialogue scenario. There exist utterance, topic and relation discrepancy mainly due to multi-speakers, utterances, and relations. In this paper, we propose a consistent learning and inference method to minimize possible contradictions from those distinctions. First, we design mask mechanisms to refine utterance-aware and speaker-aware representations respectively from the global dialogue representation for the utterance distinction. Then a gate mechanism is proposed to aggregate such bi-grained representations. Next, mutual attention mechanism is introduced to obtain the entity representation for various relation specific topic structures. Finally, the relational inference is performed through first order logic constraints over the labeled data to decrease logically contradictory predicted relations. Experimental results on two benchmark datasets show that the F1 performance improvement of the proposed method is at least 3.3% compared with SOTA.

We present a generic tool to visualize missions of the Connected Multi-Agent Path Finding (CMAPF) problem. This problem is a variant of MAPF which requires a group of agents to navigate from an initial configuration to a goal configuration while maintaining connection. The user can create an instance of CMAPF and can play the generated plan. Any algorithm for CMAPF can be plugged into the tool.

How to learn a transferable feature representation from limited examples is a key challenge for few-shot classification. Self-supervision as an auxiliary task to the main supervised few-shot task is considered to be a conceivable way to solve the problem since self-supervision can provide additional structural information easily ignored by the main task. However, learning a good representation by traditional self-supervised methods is usually dependent on large training samples. In few-shot scenarios, due to the lack of sufficient samples, these self-supervised methods might learn a biased representation, which more likely leads to the wrong guidance for the main tasks and finally causes the performance degradation. In this paper, we propose conditional self-supervised learning (CSS) to use auxiliary information to guide the representation learning of self-supervised tasks. Specifically, CSS leverages supervised information as prior knowledge to shape and improve the learning feature manifold of self-supervision without auxiliary unlabeled data, so as to reduce representation bias and mine more effective semantic information. Moreover, CSS exploits more meaningful information through supervised and the improved self-supervised learning respectively and integrates the information into a unified distribution, which can further enrich and broaden the original representation. Extensive experiments demonstrate that our proposed method without any fine-tuning can achieve a significant accuracy improvement on the few-shot classification scenarios compared to the state-of-the-art few-shot learning methods.

Hypertree width is a prominent hypergraph invariant with many algorithmic applications in constraint satisfaction and databases. We propose a novel characterization for hypertree width in terms of linear elimination orderings. We utilize this characterization to generate a new SAT encoding that we evaluate on an extensive set of benchmark instances. We compare it to state-of-the-art exact methods for computing optimal hypertree width. Our results show that the encoding based on the new characterization is not only significantly more compact than known encodings but also outperforms the other methods.

Exact cover refers to the problem of finding subfamily F of a given family of sets S whose universe is D, where F forms a partition of D. Knuth’s Algorithm DLX is a state-of-the-art method for solving exact cover problems. Since DLX’s running time depends on the cardinality of input S, it can be slow if S is large. Our proposal can improve DLX by exploiting a novel data structure, DanceDD, which extends the zero-suppressed binary decision diagram (ZDD) by adding links to enable efficient modifications of the data structure. With DanceDD, we can represent S in a compressed way and perform search in linear time with the size of the structure by using link operations. The experimental results show that our method is an order of magnitude faster when the problem is highly compressed.

This paper introduces Compositional Neural Logic Programming (CNLP), a framework that integrates neural networks and logic programming for symbolic and sub-symbolic reasoning. We adopt the idea of compositional neural networks to represent first-order logic predicates and rules. A voting backward-forward chaining algorithm is proposed for inference with both symbolic and sub-symbolic variables in an argument-retrieval style. The framework is highly flexible in that it can be constructed incrementally with new knowledge, and it also supports batch reasoning in certain cases. In the experiments, we demonstrate the advantages of CNLP in discriminative tasks and generative tasks.

Semantics based on weak admissibility were recently introduced to overcome a problem with self-defeating arguments that has not been solved for more than 25 years. The recursive definition of weak admissibility mainly relies on the notion of a reduct regarding a set E which only contains arguments which are neither in E, nor attacked by E. At first glance the reduct seems to be tailored for the weaker versions of Dung-style semantics only. In this paper we show that standard Dung semantics can be naturally reformulated using the reduct revealing that this concept is already implicit. We further identify a new abstract principle for semantics, so-called modularization describing how to obtain further extensions given an initial one. Its importance for the study of abstract argumentation semantics is shown by its ability to alternatively characterize classical and non-classical semantics.

Knowledge distillation (KD), transferring knowledge from a cumbersome teacher model to a lightweight student model, has been investigated to design efficient neural architectures. Generally, the objective function of KD is the Kullback-Leibler (KL) divergence loss between the softened probability distributions of the teacher model and the student model with the temperature scaling hyperparameter τ. Despite its widespread use, few studies have discussed how such softening influences generalization. Here, we theoretically show that the KL divergence loss focuses on the logit matching when τ increases and the label matching when τ goes to 0 and empirically show that the logit matching is positively correlated to performance improvement in general. From this observation, we consider an intuitive KD loss function, the mean squared error (MSE) between the logit vectors, so that the student model can directly learn the logit of the teacher model. The MSE loss outperforms the KL divergence loss, explained by the penultimate layer representations difference between the two losses. Furthermore, we show that sequential distillation can improve performance and that KD, using the KL divergence loss with small τ particularly, mitigates the label noise. The code to reproduce the experiments is publicly available online at https://github.com/jhoon-oh/kd_data/.

Federated Edge Learning (FEL) is a distributed Machine Learning (ML) framework for collaborative training on edge devices. FEL improves data privacy over traditional centralized ML model training by keeping data on the devices and only sending local model updates to a central coordinator for aggregation. However, challenges still remain in existing FEL architectures where there is high communication overhead between edge devices and the coordinator. In this paper, we present a working prototype of blockchain-empowered and communication-efficient FEL framework, which enhances the security and scalability towards large-scale implementation of FEL.

AlphaZero has achieved superhuman performance on various perfect-information games, such as chess, shogi and Go. However, directly applying AlphaZero to imperfect-information games (IIG) is infeasible, due to the fact that traditional MCTS methods cannot handle missing information of other players. Meanwhile, there have been several extensions of MCTS for IIGs, by implicitly or explicitly sampling a state of other players. But, due to the inability to handle private and public information well, the performance of these methods is not satisfactory. In this paper, we extend AlphaZero to multiplayer IIGs by developing a new MCTS method, Action-Prediction MCTS (AP-MCTS). In contrast to traditional MCTS extensions for IIGs, AP-MCTS first builds the search tree based on public information, adopts the policy-value network to generalize between hidden states, and finally predicts other players' actions directly. This design bypasses the inefficiency of sampling and the difficulty of predicting the state of other players. We conduct extensive experiments on the popular 3-player poker game DouDiZhu to evaluate the performance of AP-MCTS combined with the framework AlphaZero. When playing against experienced human players, AP-MCTS achieved a 65.65\% winning rate, which is almost twice the human's winning rate. When comparing with state-of-the-art DouDiZhu AIs, the Elo rating of AP-MCTS is 50 to 200 higher than them. The ablation study shows that accurate action prediction is the key to AP-MCTS winning.

The relation between Reinforcement learning (RL) and Causal Modeling(CM) is an underexplored area with untapped potential for any learning task. In this extended abstract of our Ph.D. research proposal, we present a way to combine both areas to improve their respective learning processes, especially in the context of our application area (service robotics). The preliminary results obtained so far are a good starting point for thinking about the success of our research project.

We consider the problem of the conjoint selection and allocation of projects to a population of agents, e.g. students are assigned papers and shall present them to their peers. The selection can be constrained either by quotas over subcategories of projects, or by the preferences of the agents themselves. We explore fairness and optimality issues and refine the analysis of the rank-maximality and popularity optimality concepts. We show that they are compatible with reasonable fairness requirements related to rank-based envy-freeness and can be adapted to select globally good projects according to the preferences of the agents.

Combinatorial optimization is a well-established area in operations research and computer science. Until recently, its methods have mostly focused on solving problem instances in isolation, ignoring the fact that they often stem from related data distributions in practice. However, recent years have seen a surge of interest in using machine learning, especially graph neural networks, as a key building block for combinatorial tasks, either directly as solvers or by enhancing the former. This paper presents a conceptual review of recent key advancements in this emerging field, aiming at researchers in both optimization and machine learning.

Accurate and explainable health event predictions are becoming crucial for healthcare providers to develop care plans for patients. The availability of electronic health records (EHR) has enabled machine learning advances in providing these predictions. However, many deep-learning-based methods are not satisfactory in solving several key challenges: 1) effectively utilizing disease domain knowledge; 2) collaboratively learning representations of patients and diseases; and 3) incorporating unstructured features. To address these issues, we propose a collaborative graph learning model to explore patient-disease interactions and medical domain knowledge. Our solution is able to capture structural features of both patients and diseases. The proposed model also utilizes unstructured text data by employing an attention manipulating strategy and then integrates attentive text features into a sequential learning process. We conduct extensive experiments on two important healthcare problems to show the competitive prediction performance of the proposed method compared with various state-of-the-art models. We also confirm the effectiveness of learned representations and model interpretability by a set of ablation and case studies.

Scene graphs are semantic abstraction of images that encourage visual understanding and reasoning. However, the performance of Scene Graph Generation (SGG) is unsatisfactory when faced with biased data in real-world scenarios. Conventional debiasing research mainly studies from the view of balancing data distribution or learning unbiased models and representations, ignoring the correlations among the biased classes. In this work, we analyze this problem from a novel cognition perspective: automatically building a hierarchical cognitive structure from the biased predictions and navigating that hierarchy to locate the relationships, making the tail relationships receive more attention in a coarse-to-fine mode. To this end, we propose a novel debiasing Cognition Tree (CogTree) loss for unbiased SGG. We first build a cognitive structure CogTree to organize the relationships based on the prediction of a biased SGG model. The CogTree distinguishes remarkably different relationships at first and then focuses on a small portion of easily confused ones. Then, we propose a debiasing loss specially for this cognitive structure, which supports coarse-to-fine distinction for the correct relationships. The loss is model-agnostic and consistently boosting the performance of several state-of-the-art models. The code is available at: https://github.com/CYVincent/Scene-Graph-Transformer-CogTree.