Continuous-Time Dynamic Graphs (CTDGs) enable fine-grained modeling of evolving relational systems. However, most existing CTDG representation learning methods are tailored to in-distribution settings and exhibit limited robustness under out-of-distribution (OOD) shifts. Although recent causal approaches learn invariant representations via interventions, they are primarily designed for static or discrete-time graphs and become computationally prohibitive for CTDGs due to the combinatorial explosion of structural and temporal variations. To address these challenges, we propose CIR, a framework grounded in a novel structural causal model termed the ICCM. To avoid exhaustive interventions, we leverage the Normalized Weighted Geometric Mean (NWGM) to efficiently approximate interventional predictions. We further instantiate ICCM within a practical deep learning architecture that jointly captures invariant structural and temporal patterns through dedicated subgraph extractors, and maintains an environment memory bank to model distributional shifts across evolving contexts. Extensive experiments demonstrate that CIR consistently outperforms existing methods under diverse OOD scenarios.
论文检索
输入标题、作者或关键词，从 990 篇学术成果中精准定位
Fairness-aware graph learning has become increasingly important amid growing concerns about algorithmic bias in networked data. Among existing approaches, counterfactual fairness is particularly appealing as it seeks to eliminate unfairness at its causal origin by ensuring that predictions remain invariant in counterfactual worlds where sensitive attributes are altered. However, most existing methods assume that all observed variables are directly influenced by sensitive attributes, an overly strong and often unrealistic assumption in real-world graphs. To address this limitation, we propose Graph Counterfactual Fairness (GCFair), a novel framework that achieves counterfactual fairness by explicitly identifying and disentangling the subsets of node features and graph structures genuinely affected by sensitive attributes. This principled joint disentanglement enables the generation of authentic counterfactual instances that selectively modify only sensitive-related information while preserving all sensitive-irrelevant factors. Extensive experiments show that GCFair effectively mitigates bias and outperforms state-of-the-art fairness methods in both counterfactual fairness and predictive accuracy.
Understanding underlying agent attributes such as goals, preferences, beliefs, and ability level is key to explaining decision-making and predicting behaviour in complex environments. While goal recognition (GR) has produced effective methods for inferring goals, recent work has begun to explore reinforcement learning (RL)-based approaches to attribute recognition (AR), which generalises GR by inferring a wider range of agent attributes beyond goals. Existing frameworks treat attributes as discrete variables and typically require separate models for each attribute. In reality, many agent attributes vary continuously rather than as discrete values, making continuous representations crucial for capturing subtle variations in behaviour. We introduce a multivariate RL-based AR framework that jointly infers multiple continuous-valued attributes from observed behaviour using a single attribute-conditioned policy and continuous probabilistic inference. Our experiments demonstrate that the proposed framework achieves stable and fine-grained inference, offering improved flexibility and generality compared to existing approaches.
Large Vision--Language Models (VLMs) unlearning tends to eliminate the influence of ``to-be-forgotten'' content in the training corpora, algorithmically by suppressing the likelihood of faithfully generating responses on forget-target inputs. The injection of adversarial inputs can manipulate the unlearned VLM's generation towards the attacker’s will, forcing the reproduction of the supposedly forgotten content and undermining the reliability of expected forgetting behavior. However, most attacks assume access to the unlearned VLM’s architecture or parameters, or to output logits via queries. In this paper, we propose SISA, a novel attack framework for crafting adversarial inputs to manipulate generation towards the forgotten target, which only requires access to a surrogate, pre-trained VLM. SISA advances prior attacks by exploiting the persistence of visual sink tokens after unlearning as a stable structural anchor for semantic alignment. SISA induces a sink regime on a candidate visual token to build the structural anchor that influences generation, and then semantically aligns model output to the target while conditioning on attention through the induced sink token, reinforcing the anchor for desired elicitation. With sink persistence and sink-conditioned semantic anchoring, SISA crafts transferable adversarial inputs. Evaluation on diverse unlearned VLM settings confirms the effectiveness of SISA, increasing the outputs’ semantic agreement with forgotten targets by up to 6.9X relative to clean inputs.
Ensuring the safety of probabilistic systems is a central challenge in formal verification. We propose an automated refinement technique for verifying a safety property of posterior distributions in probabilistic programs. Our approach builds on the counterexample-guided abstraction refinement (CEGAR) framework and exploits the duality in convex optimisation and the adequacy of predicate-transformer semantics in probabilistic settings. We implement the technique and evaluate its effectiveness through preliminary experiments.
Null-space-based methods have garnered considerable attention in model editing by constraining updates to the null space of the pre-existing knowledge representation, thereby preserving the model's original behavior. However, in practice these methods rely on an approximate null space—leading to knowledge leakage—and further suffer from severe performance degradation during sequential editing. Recent work shows that history-aware editing strategies can empirically mitigate this decline, yet the underlying reason remains unclear. In this paper, we first expose the knowledge leakage inherent in existing null-space approaches and then analyze why history-aware updates effectively preserve both editing performance and general capabilities during long-horizon editing. Building on these insights, we propose BetaEdit, a refined framework that effectively controls the knowledge leakage and integrates history-aware updates into the null-space paradigm. Extensive experiments on three large language models across two standard benchmarks show that BetaEdit consistently outperforms prior methods in the challenging regime of massive-scale sequential editing. Code is available at: https://github.com/lbq8942/BetaEdit.
Single-frame Infrared Small Target Detection (ISTD) aims to localize weak targets under heavy background clutter, yet dense pixel-wise annotations are expensive. Point supervision with online label evolution reduces annotation cost; however, lightweight CNN detectors often lack sufficient semantics, leading to noisy pseudo-masks and unstable optimization. To address this, we propose a hierarchical VFM-driven knowledge distillation framework that uses a frozen Vision Foundation Model (VFM) during training. We formulate point-supervised learning as a bilevel optimization process: the inner loop adapts a VFM-embedded teacher on reweighted training samples, while the outer loop transfers validation-guided knowledge to a lightweight student to mitigate pseudo-label noise and training-set bias. We further introduce Semantic-Conditioned Affine Modulation (SCAM) to inject VFM semantics into CNN features at multiple layers. In addition, a dynamic collaborative learning strategy with cluster-level sample reweighting enhances robustness to imperfect pseudo-masks. Experiments on diverse challenging cases across multiple ISTD backbones demonstrate consistent improvements in detection accuracy and training stability. Our code is available at https://github.com/yuanhang-yao/semantic-prior.
This paper explores the system 1 thinking capability of Large Reasoning Models (LRMs), the intuitive ability to respond efficiently with minimal token usage. While existing LRMs rely on long-chain reasoning and excel at complex tasks, their system 1 thinking ability remains largely underexplored. This capability is essential as it reflects models' difficulty awareness and reasoning efficiency, both critical for real-world applications. We propose S1-Bench, a multi-domain, multilingual benchmark comprising model-simple system 1 questions. Our investigation of 28 LRMs reveals under-accuracy and inefficiency on system 1 problems. We find existing efficient reasoning methods either generalize poorly to simple questions or sacrifice performance for efficiency. Further exploration uncovers LRMs' early difficulty awareness accompanied by lower confidence, and shows that problem difficulty is implicitly encoded in hidden states. Code and the extended version of the paper with appendices are available at https://github.com/WYRipple/S1_Bench.
Understanding how information or influence propagates through a network, such as during an epidemic outbreak or the spread of misinformation, is a fundamental yet challenging problem. While prior works have focused on cascade prediction (forecasting future infected nodes), network inference (recovering latent global diffusion graphs), or source localization (identifying diffusion's origin), these approaches do not recover the actual "who-infected-whom" propagation tree for a specific diffusion instance. We introduce DIPT (Deep Identification of Propagation Trees), a probabilistic framework that infers propagation trees from final observed node diffusion states, without knowledge of the underlying diffusion mechanism. DIPT models local influence strengths between nodes and uses a discrete-continuous alternating optimization strategy to jointly learn the diffusion mechanism and infer the propagation structure. Empirical results across eight real-world datasets demonstrate that DIPT consistently outperforms existing approaches in reconstructing propagation trees.
Data-free knowledge distillation enables model distillation without original training data, which is critical for privacy-sensitive tabular domains. However, existing methods do not perform well on tabular data because they do not explicitly address feature interactions, which are critical for encoding predictive knowledge. We identify interaction diversity, systematic coverage of feature combinations, as an important factor for effective tabular model distillation. To operationalize this insight, we propose TabKD, which learns adaptive feature bins aligned with teacher decision boundaries, then generates synthetic queries that ensure uniform pairwise interaction coverage. Across 4 benchmark datasets and 4 teacher architectures, TabKD achieves the highest student-teacher agreement in 14 of 16 configurations, outperforming 5 state-of-the-art baselines. We further show that interaction coverage strongly correlates with distillation quality, validating our core hypothesis. Our work establishes interaction-focused exploration as a principled framework for tabular model distillation.
Safety alignment of large language models (LLMs) has been gaining increasing attention. However, current safety-aligned LLMs suffer from the fragile and imbalanced safety mechanisms, which can still be induced to generate unsafe responses, exhibit over-safety by rejecting safe user inputs, and fail to preserve general utility after safety alignment. To this end, we propose a novel post safety alignment (PSA) method to address these inherent and emerging safety challenges, including safety enhancement, over-safety mitigation, and utility preservation. In specific, we introduce SafePatching, a novel framework for comprehensive PSA, where two distinct safety patches are developed on the harmful data to enhance safety and mitigate over-safety concerns, and then seamlessly integrated into the target LLM backbone without compromising its utility. Extensive experiments on four representative aligned LLMs, including LLaMA-2/3, Gemma and Mistral, show that SafePatching achieves a more comprehensive PSA than baseline methods, further optimizing the balance between being helpful and harmless in current aligned LLMs. Also, SafePatching demonstrates its superiority in continual PSA scenarios.
Skeleton-based human activity recognition (HAR) has achieved strong empirical performance, yet most existing models remain black boxes and difficult to interpret. In this work, we introduce a neurosymbolic formulation of skeleton-based HAR that reframes action recognition as concept-driven first-order logical reasoning over motion primitives. Our framework bridges representation learning and symbolic inference by grounding first-order logic predicates in learnable spatial and temporal motion concepts. Specifically, we employ a standard spatio-temporal skeleton encoder to extract latent motion representations, which are then mapped to interpretable concept predicates via a spatio-temporal concept decoder that explicitly separates pose-centric and dynamics-centric abstractions. These concept predicates are composed through differentiable first-order logic layers, enabling the model to learn human-readable logical rules that govern action semantics. To impose semantic structure on the learned concepts, we align skeleton representations with LLM-derived descriptions of atomic motion primitives, establishing a shared conceptual space for perception and reasoning. Extensive experiments on NTU RGB+D 60/120 and NW-UCLA demonstrate that our approach achieves competitive recognition performance while providing explicit, interpretable explanations grounded in logical structure. Our results highlight neurosymbolic reasoning as an effective paradigm for interpretable spatio-temporal action understanding.
With the increasing application of graph learning advanced by deep learning, out-of-distribution (OOD) detection for graph-structured data has become an imperative challenge in the real world. Graph neural networks (GNNs) provide a promising solution for OOD detection. However, GNNs' core message-passing mechanism inherently relies on local neighborhood aggregation, and traditional graph structures only characterize pairwise node relations, failing to capture high-order associations and global patterns. Towards this end, in this paper, we propose a Hypergraph-enhanced graph contrastive learning framework for Graph Out-Of Distribution detection (termed HGOOD). Specifically, we construct two branches to mine graph semantics in a comprehensive manner. On the one hand, we employ a graph feature branch to encode neighborhood interactions via node-level and graph-level contrastive learning. On the other hand, we incorporate the hypergraph-global branch, which adaptively models graphs’ high-order global correlations. Building upon this, we introduce a cross-branch prototype contrast that aligns the captured graph patterns with their cross-branch clustering prototypes to enhance the semantic manifold of the in-distribution (ID) graph. Extensive experiments on benchmark datasets demonstrate that our HGOOD consistently outperforms prior methods. Our code is available at https://anonymous.4open.science/r/HGOOD-CFE2/.
Spectral relaxation is widely used for large-scale combinatorial optimization due to its computational efficiency. Yet its effectiveness depends critically on the choice of graph normalization, a design decision typically made heuristically. Here, we show that normalization can be treated as a continuous optimization variable rather than a fixed preprocessing choice. Our method, Differentiable Spectral Normalization (DSN), parameterizes the spectral relaxation through a diagonal metric and maximizes the resulting lower bound via projected gradient ascent. Exact gradients are obtained through the Hellmann-Feynman theorem using only the principal eigenpair, maintaining linear complexity per iteration. On benchmark instances ranging from 10^3 to 8.4 x 10^6 nodes, DSN improves solution quality by 3-15% over static spectral methods. Its performance comes within 1-3% of state-of-the-art metaheuristics, such as simulated annealing, at up to 190x lower computational cost on large-scale instances. These results suggest that learning problem-specific relaxation geometry can substantially close the gap between spectral scalability and metaheuristic solution quality.
In many multi-agent applications, tasks yield rewards only when executed by a coalition meeting an unknown size threshold; otherwise, feedback is fully censored. This censorship creates an identifiability problem: agents cannot distinguish stochastic failure from insufficient coordination. We formalize this setting as the Threshold-Activated Cooperative Multi-Armed Bandit (TAC-MAB) and analyze it under both centralized and decentralized coordination. We show that a centralized algorithm (C-TAC) achieves cumulative regret O(log T), decomposed into a structural-search term that captures the cost of resolving feasibility under censored feedback and a statistical-monitoring term for value estimation. We then introduce D-TAC, a decentralized event-triggered protocol in which agents synchronize only when their structural beliefs change. Empirically, D-TAC achieves a 23x reduction in communication relative to the centralized baseline while preserving feasibility alignment under conservative belief fusion. These results characterize the coordination cost of learning under censored feedback and show that near-centralized communication efficiency is achievable without continuous synchronization.
Current approaches to evaluating personality in large language models (LLMs) typically prompt them to self-report on psychological questionnaires such as the Big Five Inventory. However, these methods assess introspective labels rather than observable behavior, despite the fact that LLMs are deployed to act in realistic contexts, not to reflect on their own traits. To bridge this gap, we introduce BehaviorBench, a new benchmark for evaluating LLM personality through concrete behaviors in everyday scenarios. Grounded in established personality psychology, BehaviorBench links each Big Five trait to validated behavioral manifestations and embeds them in contextually plausible situations that naturally elicit trait-relevant actions. We evaluate a range of models and personality shaping strategies using BehaviorBench and find a substantial mismatch between self-reported personalities and actual behaviors. By grounding evaluation in observable behavior rather than introspection, our work reveals critical gaps in current LLM personality modeling and control mechanisms. Our code and data are available at https://github.com/butra1n/BehaviorBench
While Vision-Language Models (VLMs) have shown promise in textual understanding, they face significant challenges when handling long context and complex reasoning tasks. In this paper, we dissect the internal mechanisms governing long-context processing in VLMs to understand their performance bottlenecks. Through the lens of attention analysis, we identify specific Visual Evidence Retrieval (VER) Heads —a sparse, dynamic set of attention heads critical for locating visual cues during reasoning, distinct from static OCR heads. We demonstrate that these heads are causal to model performance; masking them leads to significant degradation. Leveraging this discovery, we propose VERA (Visual Evidence Retrieval Augmentation), a training-free framework that detects model uncertainty (i.e., entropy) to trigger the explicit verbalization of visual evidence attended by VER heads. Comprehensive experiments demonstrate that VERA significantly improves long-context understanding of open-source VLMs: it yields an average relative improvement of 21.3% on Qwen3-VL-8B-Instruct and 20.1% on GLM-4.1V-Thinking across five benchmarks.
Multi-objective evolutionary algorithms (MOEAs) are popular tools for multi-objective optimization (MOO), and have been successfully applied to many real-world MOO problems. However, the theoretical study has lagged behind their practical success and remains largely confined to synthetic pseudo-Boolean functions. To close this gap, this paper--drawing inspiration from a class of popular continuous problems with real-world relevance--introduces a multi-objective benchmark defined on an integer space, featuring an analyzable landscape and the presence of local optima. We conduct a running time analysis on the proposed benchmark and derive several theoretical results. Specifically, we prove that a widely-studied MOEA, GSEMO, using unit-step mutation can be trapped in local optimal regions and fail to identify the Pareto front. Fortunately, we find that this difficulty can be overcome either by incorporating an ageing mechanism or using heavy-tailed mutations that allow multi-valued changes along each dimension of an individual. In addition, we demonstrate the extendability of the proposed benchmark to more complex landscapes with numerous local optima, resembling well-established problems in the field (e.g., those from the ZDT and DTLZ suites). We hope this work is a step forward for the theoretical study of MOEAs on problems that are closely related to those commonly investigated in empirical research.
Recent progress of voice conversion (VC) has achieved a new milestone in speaker cloning and linguistic preservation. But the field remains fragmented, relying on specialized models for linguistic-preserving, expressive, and singing scenarios. We propose OneVoice, a unified zero-shot framework capable of handling all three scenarios within a single model. OneVoice is built upon a continuous language model trained with VAE-free next-patch diffusion, ensuring high fidelity and efficient sequence modeling. Its core design for unification lies in a Mixture-of-Experts (MoE) designed to explicitly model shared conversion knowledge and scenario-specific expressivity. Expert selection is coordinated by a dual-path routing mechanism, including shared expert isolation and scenario-aware domain expert assignment with global-local cues. For precise conditioning, scenario-specific prosodic features are fused into each layer via a gated mechanism, allowing adaptive usage of prosody information. Furthermore, to enable the core idea and alleviate the imbalanced issue (abundant speech vs. scarce singing), we adopt a two-stage progressive training that includes foundational pre-training and scenario enhancement with LoRA-based domain experts. Experiments show that OneVoice matches or surpasses specialized models across all three scenarios, while verifying flexible control over scenarios and offering a fast decoding version as few as 2 steps. Audio samples are available at https://kerwinchao.github.io/OneVoice/.
Although Large Language Models (LLMs) excel at code generation, recent research reveals that they exhibit an insufficient grasp of core programming concepts, such as data flow and control flow. This limitation undermines their robustness when encountering variations in these concepts in practice; however, effective solutions that explicitly target this gap remain limited. To address this challenge, we propose ProCURE, a concept-aware consistency learning framework designed to enhance LLMs’ understanding of programming concepts. Specifically, ProCURE first performs automated concept-oriented code augmentation to construct a concept-aligned dataset covering representative programming concepts. It then conducts concept-aware fine-tuning, encouraging the model to capture fine-grained concept variations and learn appropriate generation behaviors under such variations via a novel concept-sensitive consistency loss. To quantify programming concept understanding, we introduce the Concept Consistency Score (CCScore), defined as the proportion of correct generations preserved under concept variations. A higher CCScore indicates a more profound understanding of programming concepts. We evaluate ProCURE on four open-source LLMs across three widely used code generation benchmarks. Experimental results show that ProCURE improves CCScore by an average of 17.9 points, demonstrating its effectiveness in addressing the programming concept understanding gap.