IJCAI 2023 Papers

Factor model is a fundamental investment tool in quantitative investment, which can be empowered by deep learning to become more flexible and efficient in practical complicated investing situations. However, it is still an open question to build a factor model that can conduct stock prediction in an online and adaptive setting, where the model can adapt itself to match the current market regime identified based on only point-in-time market information. To tackle this problem, we propose the first deep learning based online and adaptive factor model, HireVAE, at the core of which is a hierarchical latent space that embeds the underlying relationship between the market situation and stock-wise latent factors, so that HireVAE can effectively estimate useful latent factors given only historical market information and subsequently predict accurate stock returns. Across four commonly used real stock market benchmarks, the proposed HireVAE demonstrate superior performance in terms of active returns over previous methods, verifying the potential of such online and adaptive factor model.

Graph Transformer is gaining increasing attention in the field of machine learning and has demonstrated state-of-the-art performance on benchmarks for graph representation learning. However, as current implementations of Graph Transformer primarily focus on learning representations of small-scale graphs, the quadratic complexity of the global self-attention mechanism presents a challenge for full-batch training when applied to larger graphs. Additionally, conventional sampling-based methods fail to capture necessary high-level contextual information, resulting in a significant loss of performance. In this paper, we introduce the Hierarchical Scalable Graph Transformer (HSGT) as a solution to these challenges. HSGT successfully scales the Transformer architecture to node representation learning tasks on large-scale graphs, while maintaining high performance. By utilizing graph hierarchies constructed through coarsening techniques, HSGT efficiently updates and stores multi-scale information in node embeddings at different levels. Together with sampling-based training methods, HSGT effectively captures and aggregates multi-level information on the hierarchical graph using only Transformer blocks. Empirical evaluations demonstrate that HSGT achieves state-of-the-art performance on large-scale benchmarks with graphs containing millions of nodes with high efficiency.

State abstraction optimizes decision-making by ignoring irrelevant environmental information in reinforcement learning with rich observations. Nevertheless, recent approaches focus on adequate representational capacities resulting in essential information loss, affecting their performances on challenging tasks. In this article, we propose a novel mathematical Structural Information principles-based State Abstraction framework, namely SISA, from the information-theoretic perspective. Specifically, an unsupervised, adaptive hierarchical state clustering method without requiring manual assistance is presented, and meanwhile, an optimal encoding tree is generated. On each non-root tree node, a new aggregation function and condition structural entropy are designed to achieve hierarchical state abstraction and compensate for sampling-induced essential information loss in state abstraction. Empirical evaluations on a visual gridworld domain and six continuous control benchmarks demonstrate that, compared with five SOTA state abstraction approaches, SISA significantly improves mean episode reward and sample efficiency up to 18.98 and 44.44%, respectively. Besides, we experimentally show that SISA is a general framework that can be flexibly integrated with different representation-learning objectives to improve their performances further.

Weakly supervised semantic segmentation (WSSS) with image-level annotations has achieved great processes through class activation map (CAM). Since vanilla CAMs are hardly served as guidance to bridge the gap between full and weak supervision, recent studies explore semantic representations to make CAM fit for WSSS and demonstrate encouraging results. However, they generally exploit single-level semantics, which may hamper the model to learn a comprehensive semantic structure. Motivated by the prior that each image has multiple levels of semantics, we propose hierarchical semantic contrast (HSC) to ameliorate the above problem. It conducts semantic contrast from coarse-grained to fine-grained perspective, including ROI level, class level, and pixel level, making the model learn a better object pattern understanding. To further improve CAM quality, building upon HSC, we explore consistency regularization of cross supervision and develop momentum prototype learning to utilize abundant semantics across different images. Extensive studies manifest that our plug-and-play learning paradigm, HSC, can significantly boost CAM quality on both non-saliency-guided and saliency-guided baselines, and establish new state-of-the-art WSSS performance on PASCAL VOC 2012 dataset. Code is available at https://github.com/Wu0409/HSC_WSSS.

Compositional Zero-Shot Learning (CZSL) aims to imitate the powerful generalization ability of human beings to recognize novel compositions of known primitive concepts that correspond to a state and an object, e.g., purple apple. To fully capture the intra- and inter-class correlations between compositional concepts, in this paper, we propose to learn them in a hierarchical manner. Specifically, we set up three hierarchical embedding spaces that respectively model the states, the objects, and their compositions, which serve as three “experts” that can be combined in inference for more accurate predictions. We achieve this based on the recent success of large-scale pretrained vision-language models, e.g., CLIP, which provides a strong initial knowledge of image-text relationships. To better adapt this knowledge to CZSL, we propose to learn three hierarchical prompts by explicitly fixing the unrelated word tokens in the three embedding spaces. Despite its simplicity, our proposed method consistently yields superior performance over current state-of-the-art approaches on three widely-used CZSL benchmarks.

Disease progression modeling (DPM) plays an essential role in characterizing patients' historical pathways and predicting their future risks. Apprenticeship learning (AL) aims to induce decision-making policies by observing and imitating expert behaviors. In this paper, we investigate the incorporation of AL-derived patterns into DPM, utilizing a Time-aware Hierarchical EM Energy-based Subsequence (THEMES) AL approach. To the best of our knowledge, this is the first study incorporating AL-derived progressive and interventional patterns for DPM. We evaluate the efficacy of this approach in a challenging task of septic shock early prediction, and our results demonstrate that integrating the AL-derived patterns significantly enhances the performance of DPM.

In the multi-objective shortest-path problem we are interested in computing a path, or a set of paths that simultaneously balance multiple cost functions. This problem is important for a diverse range of applications such as transporting hazardous materials considering travel distance and risk. This family of problems is not new with results dating back to the 1970's. Nevertheless, the significant progress made in the field of heuristic search resulted in a new and growing interest in the sub-field of multi-objective search. Consequently, in this paper we review the fundamental problems and techniques common to most algorithms and provide a general overview of the field. We then continue to describe recent work with an emphasis on new challenges that emerged and the resulting research opportunities.

In many real-world settings, an autonomous agent may not have sufficient information or sensory capabilities to accomplish its goals, even when they are achievable. In some cases, the needed information can be provided by another agent, but information sharing might be costly due to limited communication bandwidth and other constraints. We address the problem of Helpful Information Sharing (HIS), which focuses on selecting minimal information to reveal to a partially informed agent in order to guarantee it can achieve its goal. We offer a novel compilation of HIS to a classical planning problem, which can be solved efficiently by any off-the-shelf planner. We provide guarantees of optimality for our approach and describe its extensions to maximize robustness and support settings in which the agent needs to decide which sensors to deploy in the environment. We demonstrate the power of our approaches on a set of standard benchmarks as well as on a novel benchmark.

Hawkes processes are a popular framework to model the occurrence of sequential events, i.e., occurrence dynamics, in several fields such as social diffusion. In real-world scenarios, the inter-arrival time among events is irregular. However, existing neural network-based Hawkes process models not only i) fail to capture such complicated irregular dynamics, but also ii) resort to heuristics to calculate the log-likelihood of events since they are mostly based on neural networks designed for regular discrete inputs. To this end, we present the concept of Hawkes process based on controlled differential equations (HP-CDE), by adopting the neural controlled differential equation (neural CDE) technology which is an analogue to continuous RNNs. Since HP-CDE continuously reads data, i) irregular time-series datasets can be properly treated preserving their uneven temporal spaces, and ii) the log-likelihood can be exactly computed. Moreover, as both Hawkes processes and neural CDEs are first developed to model complicated human behavioral dynamics, neural CDE-based Hawkes processes are successful in modeling such occurrence dynamics. In our experiments with 4 real-world datasets, our method outperforms existing methods by non-trivial margins.

Digraph Representation Learning aims to learn representations for directed homogeneous graphs (digraphs). Prior work is largely constrained or has poor generalizability across tasks. Most Graph Neural Networks exhibit poor performance on digraphs due to the neglect of modeling neighborhoods and preserving asymmetry. In this paper, we address these notable challenges by leveraging hyperbolic collaborative learning from multi-ordered partitioned neighborhoods and asymmetry-preserving regularizers. Our resulting formalism, Digraph Hyperbolic Networks (D-HYPR), is versatile for multiple tasks including node classification, link presence prediction, and link property prediction. The efficacy of D-HYPR was meticulously examined against 21 previous techniques, using 8 real-world digraph datasets. D-HYPR statistically significantly outperforms the current state of the art. We release our code at https://github. com/hongluzhou/dhypr.

The learnware paradigm proposed by Zhou [2016] devotes to constructing a market of numerous well-performed models, enabling users to solve problems by reusing existing efforts rather than starting from scratch. A learnware comprises a trained model and the specification which enables the model to be adequately identified according to the user's requirement. Previous studies concentrated on the homogeneous case where models share the same feature space based on Reduced Kernel Mean Embedding (RKME) specification. However, in real-world scenarios, models are typically constructed from different feature spaces. If such a scenario can be handled by the market, all models built for a particular task even with different feature spaces can be identified and reused for a new user task. Generally, this problem would be easier if there were additional auxiliary data connecting different feature spaces, however, obtaining such data in reality is challenging. In this paper, we present a general framework for accommodating heterogeneous learnwares without requiring additional auxiliary data. The key idea is to utilize the submitted RKME specifications to establish the relationship between different feature spaces. Additionally, we give a matrix factorization-based implementation and propose the overall procedure for constructing and exploiting the heterogeneous learnware market. Experiments on real-world tasks validate the efficacy of our method.

In two-player zero-sum games on graphs, the protagonist tries to achieve an objective while the antagonist aims to prevent it. Objectives for which both players do not need to use memory to play optimally are well-understood and characterized both in finite and infinite graphs. Less is known about the larger class of half-positional objectives, i.e., those for which the protagonist does not need memory (but for which the antagonist might). In particular, no characterization of half-positionality is known for the central class of ω-regular objectives. Here, we characterize objectives recognizable by deterministic Büchi automata (a class of ω-regular objectives) that are half-positional, both over finite and infinite graphs. This characterization yields a polynomial-time algorithm to decide half-positionality of an objective recognized by a given deterministic Büchi automaton.

We give polynomial time algorithms for escaping from high-dimensional saddle points under a moderate number of constraints. Given gradient access to a smooth function, we show that (noisy) gradient descent methods can escape from saddle points under a logarithmic number of inequality constraints. While analogous results exist for unconstrained and equality-constrained problems, we make progress on the major open question of convergence to second-order stationary points in the case of inequality constraints, without reliance on NP-oracles or altering the definitions to only account for certain constraints. Our results hold for both regular and stochastic gradient descent.

In this paper, we propose an HOI-aware adaptive network named AdaAct for weakly-supervised action segmentation. Most existing methods learn a fixed network to predict the action of each frame with the neighboring frames. However, this would result in ambiguity when estimating similar actions, such as pouring juice and pouring coffee. To address this, we aim to exploit temporally global but spatially local human-object interactions (HOI) as video-level prior knowledge for action segmentation. The long-term HOI sequence provides crucial contextual information to distinguish ambiguous actions, where our network dynamically adapts to the given HOI sequence at test time. More specifically, we first design a video HOI encoder that extracts, selects, and integrates the most representative HOI throughout the video. Then, we propose a two-branch HyperNetwork to learn an adaptive temporal encoder, which automatically adjusts the parameters based on the HOI information of various videos on the fly. Extensive experiments on two widely-used datasets including Breakfast and 50Salads demonstrate the effectiveness of our method under different evaluation metrics.

Human pose estimation is a challenging task due to its structured data sequence nature. Existing methods primarily focus on pair-wise interaction of body joints, which is insufficient for scenarios involving overlapping joints and rapidly changing poses. To overcome these issues, we introduce a novel approach, the High-order Directed Transformer (HDFormer), which leverages high-order bone and joint relationships for improved pose estimation. Specifically, HDFormer incorporates both self-attention and high-order attention to formulate a multi-order attention module. This module facilitates first-order "joint-joint", second-order "bone-joint", and high-order "hyperbone-joint" interactions, effectively addressing issues in complex and occlusion-heavy situations. In addition, modern CNN techniques are integrated into the transformer-based architecture, balancing the trade-off between performance and efficiency. HDFormer significantly outperforms state-of-the-art (SOTA) models on Human3.6M and MPI-INF-3DHP datasets, requiring only 1/10 of the parameters and significantly lower computational costs. Moreover, HDFormer demonstrates broad real-world applicability, enabling real-time, accurate 3D pose estimation. The source code is in https://github.com/hyer/HDFormer.

Most existing ultra-high resolution (UHR) segmentation methods always struggle in the dilemma of balancing memory cost and local characterization accuracy, which are both taken into account in our proposed Guided Patch-Grouping Wavelet Transformer (GPWFormer) that achieves impressive performances. In this work, GPWFormer is a Transformer (T)-CNN (C) mutual leaning framework, where T takes the whole UHR image as input and harvests both local details and fine-grained long-range contextual dependencies, while C takes downsampled image as input for learning the category-wise deep context. For the sake of high inference speed and low computation complexity, T partitions the original UHR image into patches and groups them dynamically, then learns the low-level local details with the lightweight multi-head Wavelet Transformer (WFormer) network. Meanwhile, the fine-grained long-range contextual dependencies are also captured during this process, since patches that are far away in the spatial domain can also be assigned to the same group. In addition, masks produced by C are utilized to guide the patch grouping process, providing a heuristics decision. Moreover, the congruence constraints between the two branches are also exploited to maintain the spatial consistency among the patches. Overall, we stack the multi-stage process in a pyramid way. Experiments show that GPWFormer outperforms the existing methods with significant improvements on five benchmark datasets.

Reinforcement learning (RL) has achieved considerable success in many fields, but applying it to real-world problems can be costly and risky because it requires a lot of online interaction. Recently, offline RL has shown the possibility of extracting a solution through existing logged data without online interaction. In this work, we propose an offline hierarchical RL method, Guider (Guide to Control), that can efficiently solve long-horizon and sparse-reward tasks from offline data. The high-level policy sequentially generates a subgoal that can guide the agent to arrive at the final goal, and the lower-level policy learns how to reach each given guided subgoal. In the process of learning from offline data, the key is to make the low-level policy reachable to the generated subgoals. We show that high-quality subgoal generation is possible through pre-training a latent subgoal prior model. The well-regulated subgoal generation improves performance while avoiding distributional shifts in offline RL by breaking down long, complex tasks into shorter, easier ones. For evaluations, Guider outperforms prior offline RL methods in long-horizon robot navigation and complex manipulation benchmarks. Our code is available at https://github.com/gckor/Guider.

As a fundamental problem in Operations Research, sparse process flexibility design (SPFD) aims to design a manufacturing network across industries that achieves a trade-off between the efficiency and robustness of supply chains. In this study, we propose a novel solution to this problem with the help of computational optimal transport techniques. Given a set of supply-demand pairs, we formulate the SPFD task approximately as a group sparse optimal transport (GSOT) problem, in which a group of couplings between the supplies and demands is optimized with a group sparse regularizer. We solve this optimization problem via an algorithmic framework of alternating direction method of multipliers (ADMM), in which the target network topology is updated by soft-thresholding shrinkage, and the couplings of the OT problems are updated via a smooth OT algorithm in parallel. This optimization algorithm has guaranteed convergence and provides a generalized framework for the SPFD task, which is applicable regardless of whether the supplies and demands are balanced. Experiments show that our GSOT-based method can outperform representative heuristic methods in various SPFD tasks. Additionally, when implementing the GSOT method, the proposed ADMM-based optimization algorithm is comparable or superior to the commercial software Gurobi. The code is available at https://github.com/Dixin-s-Lab/GSOT.

Kidney exchange programs (KEPs) typically seek to match incompatible patient-donor pairs based on a utilitarian objective where the number or overall quality of transplants is maximized---implicitly penalizing certain classes of difficult to match (e.g., highly-sensitized) patients. Prioritizing the welfare of highly-sensitized (hard-to-match) patients has been studied as a natural \textit{fairness} criterion. We formulate the KEP problem as $k$-set packing with a probabilistic group fairness notion of proportionality fairness---namely, fair $k$-set packing (\f{}). In this work we propose algorithms that take arbitrary proportionality vectors (i.e., policy-informed demands of how to prioritize different groups) and return a probabilistically fair solution with provable guarantees. Our main contributions are randomized algorithms as well as hardness results for \f{} variants. Additionally, the tools we introduce serve to audit the price of fairness involved in prioritizing different groups in realistic KEPs and other $k$-set packing applications. We conclude with experiments on synthetic and realistic kidney exchange \textsc{FairSP} instances.