IJCAI 2026 Papers

Explainable Recommendation (ER) aims to enhance recommendation transparency and prediction accuracy by providing faithful and persuasive explanations. However, Multi-Modal Multi-Interest Explainable Recommendation (MMER) is particularly challenging in two aspects: effectively utilizing diverse multi-modal information to construct reliable semantic evidence and precisely identifying the most relevant user interest to generate faithful explanations. Most previous methods fail to effectively exploit visual information to provide reliable evidence and primarily rely on a single unified representation of user preferences, making it difficult to distinguish diverse user interests and generate faithful explanations. To fill this gap, we propose Joint Multi-Modal Multi-Interest Profiling and Preference-Grounded Reasoning (PRIME) for solving the MMER problem. Specifically, PRIME first organizes items into interest-aware clusters to facilitate interest disentanglement. Based on these clusters, it constructs explicit multi-modal multi-interest user profiles and a comprehensive global item profile to capture fine-grained preferences and distinguish diverse user interests. Furthermore, PRIME performs preference-grounded reasoning by retrieving the most relevant user interest for each item, enabling faithful explanations and accurate predictions. Our experiments on three real-world datasets demonstrate that PRIME outperforms the state-of-the-art methods.

Time series forecasting plays a vital role in supporting decision-making across a wide range of critical applications, including energy, healthcare, and finance. Despite recent advances, forecasting accuracy remains limited due to the challenge of integrating historical numerical sequences with contextual features, which often comprise unstructured textual data. To address this challenge, we propose TokenCast, a large language model (LLM) driven framework that leverages language-based symbolic representations as a unified intermediary for context-aware time series forecasting. Specifically, TokenCast employs a discrete tokenizer to transform continuous numerical sequences into temporal tokens, enabling structural alignment with language-based inputs. To effectively bridge the semantic gap between modalities, both temporal and contextual tokens are embedded into a shared representation space via a pre-trained LLM, further optimized with generative objectives. Building upon this unified semantic space, the aligned LLM is subsequently fine-tuned in a supervised manner to predict future temporal tokens, which are then decoded back into the original numerical space. Extensive experiments on real-world datasets demonstrate the effectiveness of our framework and highlight its potential as a generative framework for context-aware time series forecasting. The code is available at https://github.com/Xiaoyu-Tao/TokenCast.

Most existing Graph Neural Networks (GNNs) rely on the node-level message passing or attention mechanisms to propagate and extract useful information. Although recent advances attempt to move beyond purely the node-level propagation by constructing high-level representations, these approaches are often constrained by pre-computed substructures or unidirectional bottom-up aggregations. Consequently, high-level structural semantics cannot effectively feed back to guide node representation learning, limiting the collaborative optimization between fine-grained features and macroscopic structural semantics. To address these limitations, we propose a novel Adaptive Dual-level Collaborative GNN (ADC-GNN) associated with an adaptive dual-level collaborative mechanism. We commence by introducing a set of global, learnable latent prototypes as high-level semantic references, and then employ a relaxed Sinkhorn algorithm to establish differentiable, non-collapsing assignments between nodes and prototypes. Based on these assignments, the ADC-GNN constructs high-level representations and enables interactions among them. We show that the ADC-GNN can inject the learned high-level information back into the node level, forming a closed-loop, bidirectional optimization process. Experiments demonstrate the superior performance of the proposed ADC-GNN on graph classification.

Brain effective connectivity (EC) characterizes directional causal interactions among brain regions. However, learning stable and directionally explicit EC networks from multimodal data remains challenging. In practice, functional magnetic resonance imaging (fMRI) and electroencephalography (EEG) differ substantially in spatio-temporal resolution and noise characteristics. Existing methods often rely on manual spatio-temporal alignment and tend to recover only partial causal structures under high noise and Bayesian equivalence classes. To address these challenges, we propose a spatio-temporal constrained Bayesian causal network for multimodal brain effective connectivity learning (STCBN-EC). First, STCBN-EC constructs an anatomically guided EEG–fMRI spatial mapping and derives a unified spatio-temporal representation through slice-level alignment and adaptive modality fusion. Then, a Bayesian causal network is employed to model nonlinear inter-regional dependencies, where uncertainty-driven surrogate scoring is used to evaluate candidate structures. Finally, multimodal representation learning and EC structure estimation are jointly optimized via a gradient-free global optimization strategy. Experiments on simulated and real EEG–fMRI datasets demonstrate that STCBN-EC outperforms state-of-the-art methods and effectively captures state-dependent directional interactions among brain regions.

Dynamic Graph Neural Networks (DyGNNs) are facing challenges with distribution shifts between training and test data that are similar but not identical. Existing DyGNNs for out-of-distribution scenarios primarily focus on discovering invariant patterns in the spatial domain, overlooking its impact on the structural properties in the spectral domain. In this paper, we propose a spectral-based graph augmentation framework designed to investigate and improve generalization behavior in dynamic graphs under distribution shifts. Specifically, we augment the input graph spectra into a mixture of shift components by maximizing the variance of spectral distance and propose an efficient approximation to reduce the computational cost brought by eigen-decomposition. Building on this, we develop a multi-encoder architecture in which each encoder targets a specific spectral shift component to generate referential representations. Our method adopts a new learning objective to encourage the model to rely on robust spectral properties for better generalization under distribution shifts. Extensive experiments on both real-world and synthetic datasets demonstrate that our proposed method significantly outperforms state-of-the-art approaches in node classification and link prediction tasks. Source codes are available at https://github.com/SSQiana/DSPA.

Multi-channel data fusion is essential for capturing comprehensive representations in complex systems. While graph convolutional networks have demonstrated remarkable efficacy, existing fusion paradigms primarily rely on discrete architectures governed by first-order propagation. These models are typically confined to discrete message-passing mechanisms, which makes it difficult to characterize the continuous evolution of underlying system dynamics. To address these limitations, we propose a multi-channel Graph Continuous Network (mcGCN), a novel framework for multi-channel data fusion. By formulating the information propagation as a second-order partial differential equation on graphs, mcGCN transitions from discrete layer-wise updates to a continuous dynamical system. Specifically, mcGCN integrates a multi-channel encoding module with continuous feature dynamics to initialize and evolve the latent node representations over static graph topologies. This physical analogy enables more robust information flow and effectively mitigates the performance degradation typically associated with deep graph architectures. Extensive experiments on diverse benchmark datasets demonstrate that our method outperforms state-of-the-art baselines, validating its effectiveness and robustness.

Cognitive diagnosis aims to infer students’ concept-level mastery from exercise response logs and exercise-concept associations. Fully leveraging heterogeneous relations and modeling large mastery-difficulty variations remain challenging, especially with a single predictor. To address these challenges, we propose RMCD, a unified cognitive diagnosis model that integrates relation-aware graph learning with Mixture-of-Experts (MoE) prediction. RMCD constructs a heterogeneous relational graph over students, exercises, and concepts with multiple relation types, and learns node and edge representations simultaneously. It derives relation-strength vectors from student-concept and exercise-concept edges to distinguish relation effects and refine node representations. RMCD further introduces an MoE-based prediction head that adaptively combines multiple expert predictors to capture diverse mastery-difficulty discrepancies. Experiments on benchmark datasets demonstrate that RMCD consistently outperforms state-of-the-art cognitive diagnosis methods. Our algorithm is available at https://github.com/swu-qjw-lab/code/tree/main/RMCD.

Time series anomaly detection faces a critical challenge that different anomaly types require different detection mechanisms, yet single methods are inherently limited by their design biases. We propose FlowFuse, a multi-view ensemble framework with coupling flow-based score fusion for time series anomaly detection. FlowFuse combines four complementary detection modules spanning temporal and frequency domains, as well as clustering and reconstruction paradigms, to extract multi-perspective anomaly scores. Rather than simple averaging, an ensemble of coupling flows models the joint distribution of these multi-view scores, learning complex inter-view dependencies through invertible transformations with alternating updates between temporal and frequency scores. The ensemble naturally quantifies detection uncertainty through prediction disagreement, which can optionally guide selective supervision when labels are available. Extensive experiments across 18 diverse benchmarks show that FlowFuse achieves state-of-the-art performance, demonstrating effectiveness across multiple anomaly types.

Reliable forecasting of cross-regional terrorism fatalities is essential for early warning and security planning. However, real-world terrorism data are characterized by extreme sparsity, abrupt volatility, and weak, rapidly shifting non-Euclidean spatial relationships, posing substantial challenges for existing spatiotemporal forecasting models in learning stable multi-scale dependencies. To address these challenges, we propose a robust spatiotemporal forecasting framework, termed WBMCF. In the temporal dimension, a single-level Discrete Wavelet Transform (DWT) is introduced to stabilize temporal structures, and a Bidirectional Gated Mamba module is employed to capture stage-dependent long-range temporal dependencies; additionally, in the spatial dimension, a topology-agnostic Convolutional Feed-Forwar+d Network (ConvFFN) is incorporated to model cross-regional dependencies via implicit feature coupling, thereby eliminating reliance on fixed spatial topologies. Extensive experiments on five regional subsets of the Global Terrorism Database (GTD) demonstrate that WBMCF achieves stable and competitive forecasting performance under sparse and noisy conditions. The source code is available at https://github.com/weibaozhong/WBMCF.

Over-smoothing (OSM) and over-squashing (OSQ) are two fundamental phenomena that limit the performance of Graph Neural Networks (GNNs), yet a unified spectral-spatial understanding of these phenomena remains underexplored. In this paper, we adopt polynomial spectral filters as an analytical tool to establish a unified spectral-spatial framework for graph convolution and systematically characterize the effect of the polynomial order k on information propagation in GNNs. Within this framework, we reveal an intrinsic trade-off induced by the polynomial order. Specifically, higher-order filters enhance spectral expressiveness and alleviate OSM caused by the dominant low-frequency components. However, they also expand the spatial receptive field, thereby intensifying information compression and increasing the risk of OSQ. Based on this analysis, we provide a principled guideline for selecting the polynomial order and propose a Quadratic Spectral Graph Convolution Network (QS-GCN) for graph classification. Experiments demonstrate the effectiveness and robustness of the proposed method.

Inductive knowledge graph reasoning (IKGR) requires generalization to entities and relation types not seen during training. We introduce Knowledge Graph Wave Networks (KGWN), an IKGR approach that learns adaptive per-relation wave operators to control relation-dependent propagation, along with velocity normalization ensuring stability at any propagation depth. Experiments demonstrate state-of-the-art results on inductive benchmarks and competitive transductive performance.

Text-attributed graph fraud detection (TAGFD) plays a critical role in preventing fraudulent activities on online social and e-commerce platforms. However, to evade detection, fraudsters continuously evolve their camouflaging strategies by deliberately mimicking textual responses of benign users, thereby concealing their malicious purposes. This phenomenon, referred to as semantic camouflage, fundamentally undermines commonly relied assumptions on how structural and attribute cues can be exploited to identify fraudsters, and makes it difficult to spot fraudsters with unsupervised TAGFD. To bridge the gaps, we propose a Case-Adaptive Multi-cue Expert fRAmework (CAMERA) for unsupervised TAGFD. CAMERA employs an ego-decoupled mixture-of-experts architecture, where each expert specializes in modeling a distinct type of fraud-indicative cue. A context-informed gating model is introduced to jointly consider the ego node representation and its local neighborhood context for adaptive integration of cues learned by different experts. Furthermore, CAMERA leverages the inherent rarity of fraudsters to support unsupervised one-class learning with expert-level objectives that encourage modeling dominant benign patterns, thereby enabling reliable unsupervised detection of camouflaged fraudsters. Experiments on 4 challenging datasets show that CAMERA consistently outperforms competitors, showing its effectiveness against semantically camouflaged fraudsters. Code available at https://github.com/CampanulaBells/CAMERA

Time series forecasting (TSF) plays a vital role across various domains such as finance, energy, healthcare, and meteorology. Currently, most deep learning based TSF methods typically operate with a fixed lookback window. This approach comes from the high compute and memory costs of long contexts, as well as the standard practice of using sliding windows. This creates a trade-off. Making the window larger reduces the number of training samples, which can harm stability and generalization. However, keeping the window small prevents the model from using long history during inference. We propose an inference-only streaming autoregressive framework that replaces repeated full-context recomputation with a one-time context warmup and incremental decoding, enabling efficient long-history forecasting without retraining. While straightforward caching attentions is brittle for time series due to distribution shifts and noisy or redundant histories, we address these issues with cache-consistent normalization and selective memory under a fixed cache budget. Across diverse benchmarks, our approach substantially reduces inference latency with no or marginal accuracy loss, and often improves performance when longer lookbacks are beneficial.

Social recommendation leverages social relations to enhance user preference modeling. However, real-world social networks are often noisy and unreliable, where misleading social relationships introduce anisotropic perturbations into user representations. Existing denoising approaches, including heuristic filtering and generative reconstruction, struggle to produce social embeddings that are well aligned with user preferences, limiting their effectiveness in downstream recommendation tasks. To address this challenge, we propose Manifold-Aligned Rectification Flow (MARF), a flow matching based framework that explicitly rectifies noisy social representations toward preference-aligned embeddings. MARF jointly integrates social relations and user-item interactions to construct a preference-aware manifold as the target distribution, guiding the learning of a continuous vector field that captures preference-oriented transformations in the social domain. Through this learned transport process, MARF effectively bridges the gap between the social and preference domains, yielding robust and discriminative user representations. Extensive experiments demonstrate that our proposed model consistently outperforms state-of-the-art social recommendation methods, particularly under sparse and noisy conditions, validating its effectiveness and robustness.

Driver gaze estimation serves as a fundamental metric for evaluating driver attentiveness in modern monitoring systems. Beyond being vulnerable to sudden lighting changes and sensor noise, spatial-domain models struggle to disentangle authentic gaze cues from irrelevant visual attributes. In this paper, we propose LISA, a Language-guided Interference-aware Spatial-Frequency Attention framework that combines frequency-domain priors with vision-language knowledge. Observing that the amplitude spectrum remains relatively stable even under spatial perturbations, we design a dual-domain fusion mechanism. It integrates stable low-frequency semantics into high-frequency details, employing spatial attention to precisely target ocular regions. To reduce semantic ambiguity, we also introduce a training-time disentanglement strategy. Using a frozen CLIP encoder and orthogonal regularization, we explicitly separate gaze features from appearance interference. Experiments on two benchmarks show that LISA achieves state-of-the-art performance, with significantly improved robustness against occlusions and lighting variations. The code repository is available at https://github.com/Mason-bupt/LISA.

Trajectory generation is a pivotal technique for mitigating data sparsity, but existing methods struggle to simultaneously achieve strict road network alignment and capture realistic movement characteristics. To bridge this gap, we propose RNTrajGen, a two-stage fine-grained trajectory generation framework constrained by road networks. Specifically, we first develop a Road Network Knowledge-Enhanced Encoder (RNKEE) to provide semantically rich representations for trajectory generation. By leveraging graph attention networks, RNKEE encodes static prior knowledge of the road network while integrating self-supervised learning to extract latent movement patterns from real-world trajectories. Subsequently, RNTrajGen follows a two-stage generation strategy: it first generates a sequence of road segments to ensure strict topological alignment, and then infers the moving ratios of trajectory points along road segments to capture fine-grained movement characteristics. Extensive experiments on two real-world datasets demonstrate that RNTrajGen significantly outperforms state-of-the-art baselines, and the generated trajectories exhibit high utility in downstream prediction tasks. Our code is available at https://github.com/jkzh986/RNTrajGen.

Proactive scaling improves cloud resource efficiency by forecasting system-relevant indicators and dynamically provisioning resources to maximize utilization while satisfying quality requirements. Existing approaches forecast service indicators in isolation, ignore forecasting uncertainty, and scale resource types independently, violating bundled resource constraints and degrading service quality. Therefore, we propose JointScaler, a learning-based framework for multi-indicator distribution forecasting and uncertainty-aware scaling. It captures inter-indicator dependencies via hierarchical attention, models dynamic uncertainties with normalizing flows, and leverages full predictive distributions to optimize bundled resource allocations under quality requirements. Evaluated on 4 real-world datasets, JointScaler improves point and distribution forecasting accuracy by 5.87% and 14.74%, outperforming 12 advanced baselines. In a week-long A/B test on a payment application’s cloud platform, it reduced GPU and CPU usage by 2,400+ and 37,000+ hours with stable service quality, delivering significant economic benefits.

Anomaly detection in dynamic graphs is essential for monitoring evolving systems such as transaction networks and online platforms. Yet existing methods remain limited in realistic edge-stream settings: snapshot-based approaches discretize continuous interactions and miss fine-grained temporal signals, while many continuous-time models rely on scarce node/edge attributes and often fail to explicitly assess whether a destination is compatible with a source’s recent context. Moreover, under extreme class imbalance and the lack of anomaly labels, unsupervised detectors frequently yield ill-defined normality criteria and ambiguous decision boundaries. We propose BAD, an unsupervised framework for anomaly detection in continuous-time dynamic graphs. BAD adopts a minimalistic design that represents nodes with learnable identity embeddings and performs pairwise compatibility modeling via cross-attention between each destination node and the source’s recent neighbors, enabling direct characterization of context-dependent deviations without requiring attributes. To obtain a principled separating boundary, BAD further integrates normalizing flows to model the distribution of normal interactions and derive likelihood-based anomaly scores. Extensive experiments on four real-world datasets demonstrate that BAD consistently performs well, highlighting its robustness under feature-scarce and label-scarce conditions.

Out-of-distribution (OOD) detection for time-series data remains comparatively underexplored compared to vision and language, with a limited principled understanding of how supervised time-series representations can be leveraged for reliable detection under distributional shifts. This work formulates time-series OOD detection as representation learning with hyperspherical embeddings, where class-conditional structure is induced by a von Mises–Fisher (vMF) likelihood–based objective on the unit sphere. The learned representation combines time- and frequency-domain views of the input signal via domain-specific encoders, integrating them into a joint embedding space for OOD detection. Detection uses distance-based scores over the learned embeddings, including k-nearest neighbors (k-NN) and Mahalanobis scores. We evaluate the approach at scale on the complete UCR and UEA time-series archives under a cross-dataset protocol. Empirical results show consistent improvements under both k-NN and Mahalanobis scoring over strong contrastive-learning and post-hoc baselines in the same setting. Code is available at https://github.com/tiiuae/hypertf-time-series-ood.