IJCAI 2024 Papers

Given clean labels, Graph Neural Networks (GNNs) have shown promising abilities for graph anomaly detection. However, real-world graphs are inevitably noisy labeled, which drastically degrades the performance of GNNs. To alleviate it, some studies follow the local consistency (a.k.a homophily) assumption to conduct neighborhood-based label noise correction, and to dense raw graphs using raw features or representations learned by poisoned labels. But for the anomaly detection task, the graph is not always homophilic but more likely to be heterophilic, which would corrupt the above assumption due to complicating connection patterns and impairing the effects of message passing. To this end, we propose a novel label noise-resistant graph learning (NRGL) framework, which facilitates robust graph learning from the perspectives of structure augmentation and fine-grained label governance. Specifically, we first present an investigation to verify that increasing graph homophily could help resist label noise. Based on the observation, an unsupervised contrastive learning paradigm is then introduced so well that it cannot only adaptively extract the dual views from the raw graph as structure augmentation, but also enhance the robustness of node representations. Next, given robust node representations, the noisy labels are divided into three candidate sets based on the small-loss criterion for fine-grained noise governance. Furthermore, a node sampler is designed to take structure importance, class frequency, and confidence score into consideration, which helps select reliable and important nodes for training. Extensive experiments on real-world datasets demonstrate the effectiveness of our method.

Counterfactual explanations (CEs) are advocated as being ideally suited to providing algorithmic recourse for subjects affected by the predictions of machine learning models. While CEs can be beneficial to affected individuals, recent work has exposed severe issues related to the robustness of state-of-the-art methods for obtaining CEs. Since a lack of robustness may compromise the validity of CEs, techniques to mitigate this risk are in order. In this survey, we review works in the rapidly growing area of robust CEs and perform an in-depth analysis of the forms of robustness they consider. We also discuss existing solutions and their limitations, providing a solid foundation for future developments.

Multi-view kernel clustering (MKC) aims to fully reveal the consistency and complementarity of multiple views in a potential Hilbert space, thereby enhancing clustering performance. The clustering results of most MKC methods are highly sensitive to the quality of the constructed kernels, as traditional methods independently compute kernel matrices for each view without fully considering complementary information across views. In previous contrastive multi-view kernel learning, the goal was to bring cross-view instances of the same sample closer during the kernel construction process while pushing apart instances across samples to achieve a comprehensive integration of cross-view information. However, its inherent drawback is the potential inappropriate amplification of distances between different instances of the same clusters (i.e., false negative pairs) during the training process, leading to a reduction in inter-class discriminability. To address this challenge, we propose a Robust Contrastive multi-view kernel Learning approach (R-CMK) against false negative pairs. It partitions negative pairs into different intervals based on distance or similarity, and for false negative pairs, reverses their optimization gradient. This effectively avoids further amplification of distances for false negative pairs while simultaneously pushing true negative pairs farther apart. We conducted comprehensive experiments on various MKC methods to validate the effectiveness of the proposed method. The code is available at https://github.com/Duo-laimi/rcmk_main.

Multi-modal 3D object detectors are dedicated to exploring secure and reliable perception systems for autonomous driving (AD). Although achieving state-of-the-art (SOTA) performance on clean benchmark datasets, they tend to overlook the complexity and harsh conditions of real-world environments. With the emergence of visual foundation models (VFMs), opportunities and challenges are presented for improving the robustness and generalization of multi-modal 3D object detection in AD. Therefore, we propose RoboFusion, a robust framework that leverages VFMs like SAM to tackle out-of-distribution (OOD) noise scenarios. We first adapt the original SAM for AD scenarios named SAM-AD. To align SAM or SAM-AD with multi-modal methods, we then introduce AD-FPN for upsampling the image features extracted by SAM. We employ wavelet decomposition to denoise the depth-guided images for further noise reduction and weather interference. At last, we employ self-attention mechanisms to adaptively reweight the fused features, enhancing informative features while suppressing excess noise. In summary, RoboFusion significantly reduces noise by leveraging the generalization and robustness of VFMs, thereby enhancing the resilience of multi-modal 3D object detection. Consequently, RoboFusion achieves SOTA performance in noisy scenarios, as demonstrated by the KITTI-C and nuScenes-C benchmarks. Code is available at https://github.com/adept-thu/RoboFusion.

In the face of economic downturns, Small and Medium-sized Enterprises (SMEs) within interconnected networked-loans are vulnerable to cascading debt crises, exacerbated by factors like social media-induced financial shocks. Traditional risk assessment models, which mainly rely on financial data, inadequately predict such crises, as evidenced by the collapse of Silicon Valley Bank in 2023. To address this issue, we developed RisQNet, a model that uses temporal graph networks to incorporate diverse risks, including real-time media influences. This approach not only advances risk prediction through news feature extraction and large language models but also enhances risk management strategies with intuitive visualization tools. Validated on a dataset with a total loan volume of USD 3 trillion, RisQNet outperforms the state-of-the-art baseline and achieves 87.1% of AUC. Our collaborative effort with financial regulators and the SME community underpins the model's development, aligning with the UN SDG 8. RisQNet represents a significant step forward in leveraging AI for financial stability, offering a promising approach to combat the propagation of debt crises in financial networks.

We present a new problem called rhythm inference. It consists in inferring the duration of each note and each rest from a partial specification. We then formulate rhythm inference as a constraint satisfaction problem and we use mixed linear programming to solve it. The solution is implemented for a language representing music scores, called abcd. Interestingly, the rhythm is inferred in real-time from partial musical indications.

Synthetic data-driven methods perform well on image rain removal task, but they still face many challenges in real rainfall scenarios due to the complexity and diversity of rainy patterns. In this paper, we propose a new generic paradigm for real image deraining from the perspective of synthesizing data covering more rainy patterns and constructing image rain removal networks with strong generalization performance. Firstly, instead of simply superimposing rain layers, we integrate various rainy patterns and design a phenomenal pipeline that incorporates multiple degradation types. Secondly, we construct a Patterns-aware Rain Removal Network (PRRN), which learns from both synthetic and real data simultaneously. In addition, to eliminate the inevitable distribution differences between synthetic and real data, we design a new Multi-representation Inter-domain Alignment Module (MIAM) in PRRN. By using multiple parallel submodules, MIAM achieves alignment of data domains in multiple feature subspaces. Based on several authoritative objective evaluation metrics, we successfully validate the effectiveness and robustness of the proposed method in real scenarios through extensive experiments carried out on five challenging real datasets.

Efforts to overcome catastrophic forgetting have primarily centered around developing more effective Continual Learning (CL) methods. In contrast, less attention was devoted to analyzing the role of network architecture design (e.g., network depth, width, and components) in contributing to CL. This paper seeks to bridge this gap between network architecture design and CL, and to present a holistic study on the impact of network architectures on CL. This work considers architecture design at the network scaling level, i.e., width and depth, and also at the network components, i.e., skip connections, global pooling layers, and down-sampling. In both cases, we first derive insights through systematically exploring how architectural designs affect CL. Then, grounded in these insights, we craft a specialized search space for CL and further propose a simple yet effective ArchCraft method to steer a CL-friendly architecture, namely, this method recrafts AlexNet/ResNet into AlexAC/ResAC. Experimental validation across various CL settings and scenarios demonstrates that improved architectures are parameter-efficient, achieving state-of-the-art performance of CL while being 86%, 61%, and 97% more compact in terms of parameters than the naive CL architecture in Task IL and Class IL. Code is available at https://github.com/byyx666/ArchCraft.

Causal discovery seeks to unveil causal relationships (represented as a so-called causal graph) from observational data. This paper investigates the complex relationship between the graph structure and the efficiency of constraint-based causal discovery algorithms. Our main contributions include (i) a near-tight characterization of which causal graphs admit a small d-separating set for each pair of vertices and thus can potentially be efficiently recovered by a constraint-based causal discovery algorithm, (ii) the explicit construction of a sequence of causal graphs on which the influential PC algorithm might need exponential time, although there is a small d-separating set between every pair of variables, and (iii) the formulation of a new causal discovery algorithm which achieves fixed-parameter running time by considering the maximum number of edge-disjoint paths between variables in the (undirected) super-structure as the parameter. A distinguishing feature of our investigation is that it is carried out within a more fine-grained model which more faithfully captures the infeasibility of performing accurate independence tests for large sets of conditioning variables.

The development of autonomous agents operating in dynamic and stochastic environments requires theories and models of how beliefs and intentions are revised while taking their interplay into account. In this paper, we initiate the study of belief and intention revision in stochastic environments, where an agent's beliefs and intentions are specified in a decidable probabilistic temporal logic. We then provide general Katsuno & Mendelzon-style representation theorems for both belief and intention revision, giving clear semantic characterizations of revision methods.

In this paper, we reveal the two sides of data augmentation: enhancements in closed-set recognition correlate with a significant decrease in open-set recognition. Through empirical investigation, we find that multi-sample-based augmentations would contribute to reducing feature discrimination, thereby diminishing the open-set criteria. Although knowledge distillation could impair the feature via imitation, the mixed feature with ambiguous semantics hinders the distillation. To this end, we propose an asymmetric distillation framework by feeding teacher model extra raw data to enlarge the benefit of teacher. Moreover, a joint mutual information loss and a selective relabel strategy are utilized to alleviate the influence of hard mixed samples. Our method successfully mitigates the decline in open-set and outperforms SOTAs by 2%~3% AUROC on the Tiny-ImageNet dataset and experiments on large-scale dataset ImageNet-21K demonstrate the generalization of our method.

Hierarchical leaf vein segmentation is a crucial but under-explored task in agricultural sciences, where analysis of the hierarchical structure of plant leaf venation can contribute to plant breeding. While current segmentation techniques rely on data-driven models, there is no publicly available dataset specifically designed for hierarchical leaf vein segmentation. To address this gap, we introduce the HierArchical Leaf Vein Segmentation (HALVS) dataset, the first public hierarchical leaf vein segmentation dataset. HALVS comprises 5,057 real-scanned high-resolution leaf images collected from three plant species: soybean, sweet cherry, and London planetree. It also includes human-annotated ground truth for three orders of leaf veins, with a total labeling effort of 83.8 person-days. Based on HALVS, we further develop a label-efficient learning paradigm that leverages partial label information, i.e. missing annotations for tertiary veins. Empirical studies are performed on HALVS, revealing new observations, challenges, and research directions on leaf vein segmentation. Our dataset and code are available at https://github.com/WeizhenLiuBioinform/ HALVS-Hierarchical-Vein-Segment.

In this paper we put forward a new approach to music captioning, the task of automatically generating natural language descriptions for songs. These descriptions are useful both for categorization and analysis, and also from an accessibility standpoint as they form an important component of closed captions for video content. Our method supplements an audio encoding with a retriever, allowing the decoder to condition on multimodal signal both from the audio of the song itself as well as a candidate caption identified by a nearest neighbor system. This lets us retain the advantages of a retrieval based approach while also allowing for the flexibility of a generative one. We evaluate this system on a dataset of 200k music-caption pairs scraped from Audiostock, a royalty-free music platform, and on MusicCaps, a dataset of 5.5k pairs. We demonstrate significant improvements over prior systems across both automatic metrics and human evaluation.

MaxSAT is an optimization version of the famous NP-complete Satisfiability problem (SAT). Algorithms for MaxSAT mainly include complete solvers and local search incomplete solvers. In many complete solvers, once a better solution is found, a Soft conflict Pseudo Boolean (SPB) constraint will be generated to enforce the algorithm to find better solutions. In many local search algorithms, clause weighting is a key technique for effectively guiding the search directions. In this paper, we propose to transfer the SPB constraint into the clause weighting system of the local search method, leading the algorithm to better solutions. We further propose an adaptive clause weighting strategy that breaks the tradition of using constant values to adjust clause weights. Based on the above methods, we propose a new local search algorithm called SPB-MaxSAT that provides new perspectives for clause weighting on MaxSAT local search solvers. Extensive experiments demonstrate the excellent performance of the proposed methods.

Graph classification benchmarks, vital for assessing and developing graph neural network (GNN) models, have recently been scrutinized, as simple methods like MLPs have demonstrated comparable performance. This leads to an important question: Do these benchmarks effectively distinguish the advancements of GNNs over other methodologies? If so, how do we quantitatively measure this effectiveness? In response, we first propose an empirical protocol based on a fair benchmarking framework to investigate the performance discrepancy between simple methods and GNNs. We further propose a novel metric to quantify the dataset effectiveness by considering both dataset complexity and model performance. To the best of our knowledge, our work is the first to thoroughly study and provide an explicit definition for dataset effectiveness in the graph learning area. Through testing across 16 real-world datasets, we found our metric to align with existing studies and intuitive assumptions. Finally, we explore the causes behind the low effectiveness of certain datasets by investigating the correlation between intrinsic graph properties and class labels, and we developed a novel technique supporting the correlation-controllable synthetic dataset generation. Our findings shed light on the current understanding of benchmark datasets, and our new platform could fuel the future evolution of graph classification benchmarks.

Open Set Domain Adaptation (OSDA) aims to transfer knowledge from a labeled source domain to an unlabeled target domain, where known classes exist across domains while unknown classes are present only in the target domain. Existing methods rely on the clustering structure to identify the unknown classes, which empirically induces a large identification error if the unknown classes are a mixture of multiple components. To break through this barrier, we formulate OSDA from the view of correlation and propose a correlation metric-based framework called Balanced Correlation Learning (BCL). BCL employs Hilbert-Schmidt Independence Criterion (HSIC) to characterize the separation between unknown and known classes, where HSIC is reformulated as the nodes’ relation on graph. By considering the label prior as variable, theoretical results are derived to analytically show a sufficient condition for desired learning direction for OSDA. Methodologically, the class-balanced HSIC is proposed to preserve domain-invariant and class-discriminative features. With the guarantee of correlation learning, the entropy-based principle can effectively identify the unknown classes via uncertainty. Empirically, extensive evaluations are conducted, where BCL achieves significant performance improvements.

Knowledge distillation has emerged as a highly effective method for bridging the representation discrepancy between large-scale models and lightweight models. Prevalent approaches involve leveraging appropriate metrics to minimize the divergence or distance between the knowledge extracted from the teacher model and the knowledge learned by the student model. Centered Kernel Alignment (CKA) is widely used to measure representation similarity and has been applied in several knowledge distillation methods. However, these methods are complex and fail to uncover the essence of CKA, thus not answering the question of how to use CKA to achieve simple and effective distillation properly. This paper first provides a theoretical perspective to illustrate the effectiveness of CKA, which decouples CKA to the upper bound of Maximum Mean Discrepancy (MMD) and a constant term. Drawing from this, we propose a novel Relation-Centered Kernel Alignment (RCKA) framework, which practically establishes a connection between CKA and MMD. Furthermore, we dynamically customize the application of CKA based on the characteristics of each task, with less computational source yet comparable performance than the previous methods. The extensive experiments on the CIFAR-100, ImageNet-1k, and MS-COCO demonstrate that our method achieves state-of-the-art performance on almost all teacher-student pairs for image classification and object detection, validating the effectiveness of our approaches. Our code is available in https://github.com/Klayand/PCKA.

Natural Language Inference (NLI) is a crucial task in natural language processing that involves determining the relationship between two sentences, typically referred to as the premise and the hypothesis. However, traditional NLI models solely rely on the semantic information inherent in independent sentences and lack relevant situational visual information, which can hinder a complete understanding of the intended meaning of the sentences due to the ambiguity and vagueness of language. To address this challenge, we propose an innovative ScenaFuse adapter that simultaneously integrates large-scale pre-trained linguistic knowledge and relevant visual information for NLI tasks. Specifically, we first design an image-sentence interaction module to incorporate visuals into the attention mechanism of the pre-trained model, allowing the two modalities to interact comprehensively. Furthermore, we introduce an image-sentence fusion module that can adaptively integrate visual information from images and semantic information from sentences. By incorporating relevant visual information and leveraging linguistic knowledge, our approach bridges the gap between language and vision, leading to improved understanding and inference capabilities in NLI tasks. Extensive benchmark experiments demonstrate that our proposed ScenaFuse, a scenario-guided approach, consistently boosts NLI performance.

Bokeh rendering for images shot with small apertures has drawn much attention in practice. Very recently people start to explore diffusion models for bokeh rendering, aiming to leverage the models' surging power of image generation. However, we can clearly observe two big issues with the images rendered by diffusion models: large fluctuation and severe color deviation. To address these issues, we propose in this paper a prior-aware sampling approach, which can adaptively control the noise scale through learned priors, and a prior-aware noise scheduling strategy, which can greatly reduce the number of inference steps without sacrificing performance. Extensive experiments show that our method can effectively alleviate the fluctuation problem of sampling results while ensuring similar color styles to the input image. In addition, our method outperforms state-of-the-art methods, sometimes even with only two steps of sampling. Our code is available at https://github.com/Loeiii/Reschedule-Diffusion-based-Bokeh-Rendering.