AAAI 2026 论文列表

Multi-Domain Multi-Task (MDMT) recommendation aims to provide personalized recommendations by leveraging information across multiple domains and tasks. However, existing methods often suffer from spurious correlations between irrelevant features and the target, leading to negative transfer. To address this, we propose a Stable and Adaptive Fusion (SAF) framework for MDMT recommendation. SAF introduces a weighted Hilbert-Schmidt Independence Criterion (HSIC) loss to decorrelate irrelevant features from the target, learning sample weights that promote stable (i.e., robust to spurious correlations) representations in both bottom and expert layers. We employ Random Fourier Features (RFF) to enable scalable computation of the HSIC loss. We further employ adaptive feature and expert gating to select these stable features, enabling the model to capture intricate cross-domain and cross-task dependencies. The learned sample weights are also used to reweight the MDMT loss during training. Experiments on large-scale datasets show that SAF outperforms state-of-the-art baselines by up to 2% in AUC. To facilitate further research, we release a new industrial dataset with 30 million interactions across 3 domains and 2 tasks, with 300 features.

Learning representation of the enclosing subgraph of node pairs is recognized as an efficient approach for link-oriented prediction tasks in network applications. The core challenge within this subgraph encoding approach is how to effectively distinguish and then properly aggregate the contribution of nodes in the subgraph into a single vector to indicate the relation between the target node pair. In this work, we propose a novel sphere-based subgraph encoding architecture, namely BS-SubGNN, to address the challenge. In detail, we design two key building blocks, including Bicentric Sphere Node Labeling (BSNL) and Bicentric Sphere Subgraph Pooling (BSSP) to assist message passing in BS-SubGNN. BSNL endows each node a label according to the sphere it belongs to in the subgraph to distinguish the contribution of nodes, while BSSP adopts an attention mechanism to aggregate the contribution of nodes in each sphere. Theoretically, we prove that BS-SubGNN can unify existing node distance labeling methods, and yield discriminative node features with less time complexity. We evaluate the performance of BS-SubGNN in link prediction tasks over a variety of network types, including undirected networks, attribute networks, directed networks, and signed directed networks. Our experimental results demonstrate that BS-SubGNN consistently achieves significant performance improvements over the above diverse types of networks. In particular, compared to those methods with a requisite of multi-hop neighborhood information, BS-SubGNN can obtain better performance even when only one-hop neighborhood information of the node pair is utilized.

User purchase decisions are driven by complex, multi-faceted intentions that evolve across different temporal horizons (e.g., immediate needs, transitional interests, and long-term preferences). However, existing sequential methods often treat user sequences as unified blocks, overlooking the dynamic evolution of user intents at different granularities, while also lacking robustness against prevalent noise in real-world interaction data. This paper proposes Multi-granularity Intent Modeling with Adversarial Robustness for Sequential Recommendation (MIMAR-SRec), a framework that models latent user intentions at multiple granularities. Specifically, MIMAR-SRec integrates multi-granularity intent representation across different contextual windows to capture evolving user interests, dual-perspective contrastive learning that aligns user representations with both intent prototypes and cross-user sequences, and intent-similarity adversarial robustness that systematically enhances model stability against interaction, temporal, and preference noise through controlled perturbations. By integrating multi-granularity intent modeling with adversarial training, MIMAR-SRec enables simultaneous fine-grained underlying intent modeling and noise-resistant recommendations. Extensive experiments on four widely used benchmark datasets demonstrate that MIMAR-SRec outperforms baselines, particularly in long-tail item recommendation and noisy interaction scenarios.

Knowledge Tracing (KT) aims to mine students’ evolving knowledge states and predict their future question-answering performance. Existing methods based on heterogeneous information networks (HINs) are prone to introducing noises due to manual or random selection of meta-paths and lack necessary quality assessment of meta-path instances. Conversely, recent large language models (LLMs)-based methods ignore the rich information across students, and both paradigms struggle to deliver consistently accurate and evidence-based explanations. To address these issues, we propose an innovative framework, HIN-LLM Synergistic Enhanced Knowledge Tracing (HISE-KT), which seamlessly integrates HINs with LLMs. HISE-KT first builds a multi-relationship HIN containing diverse node types to capture the structural relations through multiple meta-paths. The LLM is then employed to intelligently score and filter meta-path instances and retain high-quality paths, pioneering automated meta-path quality assessment. Inspired by educational psychology principles, a similar student retrieval mechanism based on meta-paths is designed to provide a more valuable context for prediction. Finally, HISE-KT uses a structured prompt to integrate the target student's history with the retrieved similar trajectories, enabling the LLM to generate not only accurate predictions but also evidence-backed, explainable analysis reports. Experiments on four public datasets show that HISE-KT outperforms existing KT baselines in both prediction performance and interpretability.

The increasing scale of graph datasets has significantly improved the performance of graph representation learning methods, but it has also introduced substantial training challenges. Graph dataset condensation techniques have emerged to compress large datasets into smaller yet information-rich datasets, while maintaining similar test performance. However, these methods strictly require downstream applications to match the original dataset and task, which often fails in cross-task and cross-domain scenarios. To address these challenges, we propose a novel causal-invariance-based and transferable graph dataset condensation method, named TGCC, providing effective and transferable condensed datasets. Specifically, to preserve domain-invariant knowledge, we first extract domain causal-invariant features from the spatial domain of the graph using causal interventions. Then, to fully capture the structural and feature information of the original graph, we perform enhanced condensation operations. Finally, through spectral-domain Enhanced contrastive learning, we inject the causal-invariant features into the condensed graph, ensuring that the compressed graph retains the causal information of the original graph. Experimental results on five public datasets and our novel FinReport dataset demonstrate that TGCC achieves up to a 13.41% improvement in cross-task and cross-domain complex scenarios compared to existing methods, and achieves state-of-the-art performance on 5 out of 6 datasets in the single dataset and task scenario.

Generative recommendation (GR) is an emerging paradigm that tokenizes items into discrete tokens and learns to autoregressively generate the next tokens as predictions. While this token-generation paradigm is expected to surpass traditional transductive methods, potentially generating new items directly based on semantics, we empirically show that GR models predominantly generate items seen during training and struggle to recommend unseen items. In this paper, we propose SpecGR, a plug-and-play framework that enables GR models to recommend new items in an inductive setting. SpecGR uses a drafter model with inductive capability to propose candidate items, which may include both existing items and new items. The GR model then acts as a verifier, accepting or rejecting candidates while retaining its strong ranking capabilities. We further introduce the guided re-drafting technique to make the proposed candidates more aligned with the outputs of generative recommendation models, improving the verification efficiency. We consider two variants for drafting: (1) using an auxiliary drafter model for better flexibility, or (2) leveraging the GR model's own encoder for parameter-efficient self-drafting. Extensive experiments on three real-world datasets demonstrate that SpecGR exhibits both strong inductive recommendation ability and the best overall performance among the compared methods.

Attributed heterogeneous information networks (AHINs) encode rich semantics through diverse node and edge types. Recent learning-based community search methods on AHINs have shown promising performance but face two major limitations: i) difficulty scaling to large graphs due to memory-intensive neighbor-based propagation (e.g., GNNs and node-level attention), and ii) reliance on explicit community-level labels, which are often unavailable or costly to obtain. To address these issues, we propose a scalable Semi-supervised Community Search framework on AHINs (SCSAH), enabling scalability and efficiency, while eliminating the need for community-level labels by leveraging readily available node classification labels. Specifically, we devise MvSF2Token to extract Multi-view Semantic Features (MvSFs) as compact subgraph-level tokens before training, significantly reducing model propagation complexity. We then design a View-Aware Semantic Graph Transformer (VASGhormer) to effectively encode MvSFs by capturing cross-view dependencies and fusing semantic features. The combination of MvSF2Token and VASGhormer ensures scalability, efficiency, and robust performance. Furthermore, we design a View-Aware Contrastive Learner to train VASGhormer without requiring community-level supervision. Extensive experiments on five real-world datasets show that SCSAH outperforms state-of-the-art methods, achieving 18.06% higher performance and 10.43 times faster training.

Inductive knowledge graph completion (KGC) aims to predict missing links involving unseen entities, making it a particularly challenging task for knowledge representation learning. Traditional embedding-based methods often fall short in this setting due to their limited structural reasoning capabilities. Recently, Graph Neural Networks (GNNs) offer a promising alternative by explicitly modeling the graph topology. However, their performance heavily relies on the quality of negative samples during training, which significantly influences the learned representations and generalization ability. To tackle this issue, we propose Adaptive Relation-Aware Negative Sampling (ARNS), a negative sampling approach specifically tailored for GNN-based inductive KGC. It integrates three key strategies: (1) High-quality negatives via Linear WD for discriminative learning, (2) Relation-aware negatives utilizing relation graphs to preserve structural patterns, as well as (3) Adaptive curriculum learning that dynamically adjusts sampling ratios based on performance feedback. Our key innovation lies in a performance-driven adaptation mechanism that monitors training dynamics and modulates negative sample difficulty. This approach starts with easier samples for stability, and progressively introduces challenging negatives. Experiments demonstrate that ARNS outperforms state-of-the-art methods with significant MRR improvements while maintaining training stability. The adaptive design is particularly beneficial in inductive scenarios, where models can infer structural patterns from limited observations.

In online advertising under the cost-per-conversion (CPA) model, accurate conversion rate (CVR) prediction is crucial. A major challenge is delayed feedback, where conversions may occur long after user interactions, leading to incomplete recent data and biased model training. Existing solutions partially mitigate this issue but often rely on auxiliary models, making them computationally inefficient and less adaptive to user interest shifts. We propose IF-DFM, an Influence Function-empowered for Delayed Feedback Modeling which estimates the impact of newly arrived and delayed conversions on model parameters, enabling efficient updates without full retraining. By reformulating the inverse Hessian-vector-product as an optimization problem, IF-DFM achieves a favorable trade-off between scalability and effectiveness. Experiments on benchmark datasets show that IF-DFM outperforms prior methods in both accuracy and adaptability.

The number of n-gram features grows exponentially in n, making it computationally demanding to compute the most frequent n-grams even for n as small as 3. Motivated by our production machine learning system built on n-gram features, we ask: is it possible to accurately, deterministically, and quickly recover the top-k most frequent n-grams? We devise a multi-pass algorithm called Intergrams that constructs candidate n-grams from the preceding (n-1)-grams. By designing this algorithm with hardware in mind, our approach yields more than an order of magnitude speedup (up to 33x!) over the next known fastest algorithm, even when similar optimization are applied to the other algorithm. Using the empirical power-law distribution over n-grams, we also provide theory to inform the efficacy of our multi-pass approach.

Sequential recommendation (SR) aims to predict users' next action based on their historical behavior, and is widely adopted by a number of platforms. The performance of SR models relies on rich interaction data. However, in real-world scenarios, many users only have a few historical interactions, leading to the problem of data sparsity. Data sparsity not only leads to model overfitting on sparse sequences, but also hinders the model’s ability to capture the underlying hierarchy of user intents. This results in misinterpreting the user's true intents and recommending irrelevant items. Existing data augmentation methods attempt to mitigate overfitting by generating relevant and varied data. However, they overlook the problem of reconstructing the user's intent hierarchy, which is lost in sparse data. Consequently, the augmented data often fails to align with the user's true intents, potentially leading to misguided recommendations. To address this, we propose the Adaptive Diffusion Augmentation for Recommendation (ADARec) framework. Critically, instead of using a diffusion model as a black-box generator, we use its entire step-wise denoising trajectory to reconstruct a user's intent hierarchy from a single sparse sequence. To ensure both efficiency and effectiveness, our framework adaptively determines the required augmentation depth for each sequence and employs a specialized mixture-of-experts architecture to decouple coarse- and fine-grained intents. Experiments show ADARec outperforms state-of-the-art methods on standard benchmarks and on sparse sequences, demonstrating its ability to reconstruct hierarchical intent representations from sparse data.

Neural video compression (NVC) has demonstrated superior compression efficiency, yet effective rate control remains a significant challenge due to complex temporal dependencies. Existing rate control schemes typically leverage frame content to capture distortion interactions, overlooking inter-frame rate dependencies arising from shifts in per-frame coding parameters. This often leads to suboptimal bitrate allocation and cascading parameter decisions. To address this, we propose a reinforcement‑learning (RL)‑based rate control framework that formulates the task as a frame‑by‑frame sequential decision process. At each frame, an RL agent observes a spatiotemporal state and selects coding parameters to optimize a long‑term reward that reflects rate‑distortion (R-D) performance and bitrate adherence. Unlike prior methods, our approach jointly determines bitrate allocation and coding configuration in a single step, independent of group‑of‑pictures (GOP) structure. Extensive experiments across diverse NVC architectures show that our method reduces the average relative bitrate error to 1.20 percent and achieves up to 13.45 percent bitrate savings at typical GOP sizes, outperforming existing approaches. In addition, our framework demonstrates improved robustness to content variation and bandwidth fluctuations with lower encoding/decoding overhead, making it highly suitable for practical deployment.

Learning path recommendation seeks to provide students with a structured sequence of learning items (e.g., knowledge concepts or exercises) to optimize their learning efficiency. Despite significant efforts in this area, most existing methods primarily rely on prerequisite relations, which present two major limitations: (1) Prerequisite relations between knowledge concepts are difficult to obtain due to the cost of expert annotation, hindering the application of current learning path recommendation methods. (2) Relying on a single sequentially dependent knowledge structure based on prerequisite relations implies that a confusing knowledge concept can disrupt subsequent learning processes, which is referred to as blocked learning. To address these two challenges, we propose a novel approach, GraphRAG-Induced Dual Knowledge Structure Graphs for Personalized Learning Path Recommendation (KnowLP), which enhances learning path recommendations by incorporating both prerequisite and similarity relations between knowledge concepts. Specifically, we introduce a knowledge structure graph generation module EDU-GraphRAG that constructs knowledge structure graphs for different educational datasets, significantly improving the applicability of learning path recommendation methods. We then propose a Discrimination Learning-driven Reinforcement Learning (DLRL) module that utilizes similarity relations as fallback relations when prerequisite relations become ineffective, thereby alleviating the blocked learning. Finally, we conduct extensive experiments on three benchmark datasets, demonstrating that our method not only achieves state-of-the-art performance but also generates more effective and longer learning paths.

Topological Data Analysis (TDA) provides artificial intelligence (AI) systems with mathematically rigorous geometric descriptors through Persistent Homology (PH), capturing essential shape characteristics in high-dimensional data. Yet, PH’s combinatorial complexity and sensitivity to outliers hinder its scalability and reliability, especially for Intrinsic PH (IPH) that relies on accurate geodesic distances. While stateof-the-art landmark-based subsampling methods, PH Landmarks, ameliorate computational costs and improve outlier robustness by selecting representative points based on local PH scores, it remain computationally intensive and at low sampling rates struggle to reconstruct the global topology. In this work, we introduce TOPOGRAPH, a simple yet powerful framework that preserves intrinsic topology. The resulting coarsened graph supports efficient IPH computations using Fermat distances. Experiments on both synthetic and realworld datasets show that TOPOGRAPH outperforms stateof-the-art sampling-based methods by achieving an order-ofmagnitude speedup and substantially improved topological fidelity in persistence diagrams, demonstrating its ability for robust and scalable topological data analysis.

Learning representations on graphs is foundational for many downstream tasks, and its synergy with diffusion models has emerged as a promising direction. However, diffusion-based methods for heterogeneous graphs remain underexplored, confronting two principal challenges: (1) The presence of noise and structural heterogeneity in graphs makes it challenging to accurately capture semantic transitions among diverse relation types. (2) The isotropic Gaussian noise used in forward diffusion fails to reflect graphs' inherent semantics and structural anisotropy. To address these, we propose ARDiff, a novel framework that integrates residual diffusion with anisotropic noise for heterogeneous graph learning. Specifically, we propose a semantic residual diffusion mechanism that progressively refines node embeddings by orchestrating transitions from low-semantic (high-noise) to high-semantic (low-noise) relational contexts, thus enabling step-wise distillation of task-relevant information. In addition, to address the limitations of conventional diffusion, we introduce an anisotropic diffusion strategy: in the forward process, noise injection is oriented by structural and semantic priors; in the denoising step, a conditional diffusion mechanism is guided by a random walk encoding, enhancing both topological consistency and semantic alignment. Extensive evaluation on heterogeneous graph datasets demonstrates that ARDiff significantly surpasses current leading methods in link prediction and node classification, setting a new paradigm and benchmark in heterogeneous graph representation learning.

Graph Contrastive Learning (GCL) has emerged as a powerful paradigm for training Graph Neural Networks (GNNs) in the absence of task-specific labels. However, its scalability on large-scale graphs is hindered by the intensive message passing mechanism of GNN and the quadratic computational complexity of contrastive loss over positive and negative node pairs. To address these issues, we propose an efficient GCL framework that transforms the input graph into a compact network of interconnected node sets while preserving structural information across communities. We firstly introduce a kernelized graph community contrastive loss with linear complexity, enabling effective information transfer among node sets to capture hierarchical structural information of the graph. We then incorporate a knowledge distillation technique into the decoupled GNN architecture to accelerate inference while maintaining strong generalization performance. Extensive experiments on sixteen real-world datasets of varying scales demonstrate that our method outperforms state-of-the-art GCL baselines in both effectiveness and scalability.

Traditional short video recommendations primarily enhance user retention by reinforcing existing user preferences, potentially leading to information cocoons. Conversely, proactive recommendations aim to diversify user interests by exposing users to content beyond their historical preferences. However, current proactive approaches face three limitations: (1) homogeneous receptivity assumption, neglecting individual differences in users' openness to new interests; (2) short-term item exposure without interest anchoring, focusing on item-level shifts rather than interest evolution; and (3) static feedback utilization, failing to incorporate dynamic user feedback during the recommendation adequately. To address these challenges, we propose ProRec-Video, a proactive framework that guides hierarchical interest transitions through three innovations. First, User Receptivity Profiling assesses individual openness for new interests, ensuring personalized transition pacing. Second, Hierarchical Interest Transition Planning decomposes complex interest shifts into intermediate steps to generate smooth interest transition paths and semantically coherent video sequences, addressing overemphasis on item exposure. Third, Dynamic Feedback Adaptation integrates agent-based simulation and Reflexion mechanisms to refine interest transition paths and video sequences based on real-time user feedback, enhancing adaptability and satisfaction. Extensive experiments on two datasets demonstrate that ProRec-Video achieves a significant improvement in proactive recommendation performance, with an interest transition success rate of 85% and a user satisfaction rate of 78.3%.

Feature coding has recently emerged as a key technique for efficient transmission of intermediate representations in distributed AI systems. Existing approaches largely follow a transform-based pipeline inherited from image and video coding, where the transform module is used to remove spatial structural redundancies in visual signals. However, our analysis indicates that such redundancies have already been largely removed during feature extraction, which reduces the necessity of the transform module. Building on this insight, we propose a new transform-free pipeline that directly encodes the extracted features via a vector quantization module and an entropy model. The proposed transform‑free framework jointly learns the quantization codebook and entropy model, enabling end‑to‑end optimization tailored to the inherent feature characteristics. Furthermore, the proposed method inherently avoids the computational complexity of the transform module. Experiments on features from diverse architectures and tasks demonstrate that our method achieves superior rate-distortion performance compared to transform-based baselines, while significantly reducing the encoding and decoding complexity.

Existing core-set selection methods predominantly rely on heuristic scoring signals such as training dynamics or model uncertainty, lacking explicit modeling of data likelihood. This omission may hinder the constructed subset from capturing subtle yet critical distributional structures that underpin effective model training. In this work, we propose a novel, theoretically grounded approach that leverages diffusion models to estimate data likelihood via reconstruction deviation induced by partial reverse denoising. Specifically, we establish a formal connection between reconstruction error and data likelihood, grounded in the Evidence Lower Bound (ELBO) of Markovian diffusion processes, thereby enabling a principled, distribution-aware scoring criterion for data selection. Complementarily, we introduce an efficient information-theoretic method to identify the optimal reconstruction timestep, ensuring that the deviation provides a reliable signal indicative of underlying data likelihood. Extensive experiments on ImageNet demonstrate that reconstruction deviation offers an effective scoring criterion, consistently outperforming existing baselines across selection ratios, and closely matching full-data training using only 50% of the data. Further analysis shows that the likelihood-informed nature of our score reveals informative insights in data selection, shedding light on the interplay between data distributional characteristics and model learning preferences.