Parameter sharing is a central design choice in cooperative multi-agent reinforcement learning, yet it fundamentally conflicts with the need for role specialization in heterogeneous cooperative environments. Existing role-based methods typically learn monolithic role representations, which often suffer from gradient interference and fail to capture the compositional structure of complex behaviors. Inspired by Trait Theory, we propose DEcompose and COnstruct Roles (DECOR), a framework that models agent roles as dynamic compositions of orthogonal behavioral traits. DECOR introduces an orthogonal Mixture-of-Experts architecture to decompose behaviors into independent traits, mitigating destructive gradient interference under parameter sharing, and a group-consensus guided mechanism to extract team-level tactical intents that guide role composition.Experiments on multiple benchmarks demonstrate that DECOR consistently improves sample efficiency and overall performance over existing related methods.
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
Joint optimization of pricing, dispatching, and routing is critical for hub-based mobility services but challenging due to complex decision couplings and strict service guarantees, such as Order Response Rate (ORR). Conventional constrained reinforcement learning often struggles in this mixed continuous--combinatorial action space, suffering from oscillatory behavior in Lagrangian dual variables and unstable constraint satisfaction. To address this, we propose PID-SACA, a unified framework that integrates an entropy-regularized actor--critic policy for continuous pricing and dispatching assisted by an embedded routing solver for execution-aware feedback. Crucially, we adapt the PID control mechanism to the Lagrangian dual update process. This approach leverages proportional, integral, and derivative feedback to dampen oscillations caused by stochastic gradient variance, ensuring robust long-term constraint enforcement. We provide theoretical analysis on the boundedness of dual variables, and experiments on publicly available large-scale mobility datasets demonstrate that PID-SACA significantly outperforms baselines, achieving high revenue with stable service compliance. Code: https://github.com/jerry0375/PID-SACA
Multi-view clustering aims to discover shared semantic structures from multiple complementary data views to improve clustering performance. However, most existing methods rely on a single clustering representation for each semantic cluster, which limits the ability to model complex cluster structures and diverse patterns in real-world data. To address this limitation, we propose a Mixture of Clustering Experts with Dual Consistency (MoCE-DC). MoCE-DC introduces a set of learnable clustering experts that characterize the same semantic cluster from multiple perspectives within a unified prototype space. A gating mechanism is employed to adaptively select experts for each sample. MoCE-DC decomposes complex semantic clusters into multiple collaborative sub-structures, significantly enhances the ability to model intricate intra-cluster diversity. To further ensure cross-view semantic consistency, we propose a dual-level alignment mechanism that enforces prediction consistency across views while guiding clustering assignments toward a more discriminative direction. In addition, a gating balance regularization strategy is introduced to mitigate expert collapse and promote balanced expert utilization. Extensive experiments on multiple public multi-view clustering benchmarks demonstrate that MoCE-DC outperforms state-of-the-art methods.
Agents implemented using Large Language Models (LLMs) offer novel solutions for tasks such as document summarization. However, single-agent systems, limited by their own knowledge base, often exhibit information omissions or inconsistencies with the facts, while multi-agent systems lacking effective organization can fall into the trap of social cognitive defects such as flattery and premature consensus. Inspired by the U.S. jury system, we propose JURIS, a judicial-decision-inspired multi-agent collaborative framework. Specifically, the document summarization process simulates a lawsuit scenario: multiple generator agents condense the target text into multiple sentences from different perspectives, much like lawyers on both sides providing different evidence; multiple decision-making agents consider the overall picture based on defense and voting mechanisms to select candidate sentences, much like a jury adopting different evidence; finally, a chief editor agent compiles a summary of all selected sentences, much like a secretary summarizing a plan and a judge giving a final verdict. Extensive experiments on in-domain news and cross-domain benchmarks demonstrate that JURIS consistently and significantly outperforms strong single-agent and multi-agent baselines across automatic evaluation metrics, multi-dimensional quality assessments, and human judgments, validating the effectiveness of decision-driven structured multi-agent collaboration for document summarization.
Flow Matching (FM), as a highly promising generative paradigm, has recently been introduced into sequential recommendation. However, the lack of explicit inter-user supervision in existing FM-based method may lead to an averaged flow phenomenon, where the model tends to produce similar velocity directions for different users during item generation. This greatly undermines the model’s ability to generate personalized item. To address this issue, we propose a Contrastive Flow Matching (ConFM) for sequential reconmendation, aiming to inject inter-user separation signals so as to preserve the discriminability of users’ velocity directions. Specifically, considering the common design where velocities are derived from predicted item representations, we introduce a Representation Contrastive Regularizer in the representation space, which forces the predicted item embeddings of different users to be separated from each other. In addition, we introduce a Velocity Contrastive Regularizer in the induced velocity space, which directly encourages each user’s induced velocity to stay far from other users’ velocity targets, thereby better aligning the generation process with the user’s own interests. Extensive experiments on four datasets demonstrate that ConFM consistently outperforms strong baselines, and it also achieves notable improvements on long-tail items.
When vision contradicts text, multimodal large language models (MLLMs) consistently favor text—even when images provide clear evidence otherwise. This bias poses risks for applications requiring visual grounding, yet its cause remains unclear. In this paper, we uncover a surprising finding: models often get it right initially, forming correct vision-based predictions in their intermediate layers, before changing their minds and favoring text in the final output. We call this "late-layer textual override". The visual information is encoded, it simply does not survive to the output. More intriguingly, we find that how predictions change reveals whether they're correct: 85% of failures shift toward text, while 89% of successes shift toward vision. This directional signature enables a simple but powerful intervention: when we detect a confident visual prediction being suppressed, we restore it. We propose CALRD (Conflict-Aware Layer Reference Decoding), a training-free method that recovers overridden predictions at inference time. Experiments across five MLLMs of varying architectures demonstrate up to 9.4% absolute improvements on conflict benchmarks while largely preserving standard performance, without training or external knowledge. It recovers what the model already knew but failed to preserve.
Molecular property prediction is a fundamental task with wide-ranging applications in chemistry, materials science, and drug discovery. Beyond predictive accuracy, the reliability of model predictions, particularly their uncertainty awareness, is critical for high-stakes downstream applications. However, most existing methods predominantly optimize performance metrics while overlooking prediction reliability, often producing results that are both inaccurate and overconfident. To address this challenge, we propose an uncertainty-guided framework that integrates adaptive local adversarial perturbation with evidential deep learning for reliable molecular property prediction. The proposed method identifies sensitive molecular regions through gradient-based sensitivity analysis and applies locally adaptive perturbations whose magnitudes are dynamically modulated by predictive uncertainty. This approach explicitly challenges the model to preserve predictive stability under structure-aware perturbations, thereby enhancing its sensitivity to critical molecular substructures. Furthermore, we introduce an evidential calibration strategy that aligns uncertainty with the prediction discrepancy between clean and adversarial molecular views, enabling the learned evidence to faithfully reflect prediction stability. Extensive experiments on benchmark molecular property prediction datasets demonstrate that our approach achieves state-of-the-art predictive performance while providing more reliable uncertainty estimates than widely used uncertainty quantification methods. These results highlight the potential of the proposed framework as a reliable tool for risk-sensitive virtual screening and molecular decision-making. Our code is available at https://github.com/Ubehind/ugap.
Generating protein sequences that reliably fold into target structures is a central challenge in computational biology and protein design. Progress in protein inverse folding (PIF), however, is fundamentally constrained by the scarcity of high-quality structural data, which limits the effectiveness of existing models. In this work, we propose TTS-Design, a test-time compute scaling framework that enhances protein sequence design without retraining models or relying on larger training data. For TTS-Design, its core design is a dedicated inverse folding reward model named IF-RM, which quantitatively evaluates structure–sequence compatibility and enables effective selection among multiple candidate sequences at inference time. By integrating IF-RM into existing PIF models, TTS-Design leverages additional computation during inference stage to explore and refine candidate solutions. Extensive experiments on the CATH 4.2 and CATH 4.3 demonstrate that TTS-Design consistently improves sequence recovery and structural reliability across different backbone models, without retraining or increasing model size. Case studies on real-world proteins further confirm its practical effectiveness. Overall, our results highlight test-time compute scaling as a promising and effective paradigm for advancing protein inverse folding and other scientific problems under data-limited scenarios.
Temporal link prediction with temporal graph neural networks (TGNNs) is increasingly used to model spatio-temporal dependencies in temporal graphs and to forecast future interactions among entities. Existing sampling-based training methods typically rely on random negative sampling and pointwise loss formulations, which often lead to suboptimal convergence and limited generalization due to low-quality negative samples. We propose ATNSF, a temporal graph learning framework with a hybrid negative sampling strategy that uses a portion of historical edges as hard negatives. For efficiency, we design an asynchronous parallel training pipeline for scalable optimization and introduce a pairwise sampled softmax loss that contrasts each positive instance with a batch of negatives to learn more discriminative representations. Finally, we theoretically show that jointly designing the loss function and negative sampling strategy is crucial for improving performance and generalization. Extensive experiments across six temporal graph datasets demonstrate that ATNSF improves the average AP from 0.724 to 0.827 (+0.103). Remarkably, it also accelerates training by 1.37× to 6.85×, achieving a 2.57× geometric mean speedup. The source code of this paper can be found at https://github.com/yongqiu-star/ATNSF.
Bound-Independent Pruning (BIP) is a powerful technique for accelerating tree-based complete search algorithms for Distributed Constraint Optimization Problems (DCOPs) by pruning dominated regions of the search space using only local information. However, BIP fundamentally relies on constructing a context-dependent dominating space through joint optimization over local constraints, incurring exponential computational and storage complexity that limits its practical scalability. To address this limitation, we propose DR-BIP, a decoupled and sound reformulation of BIP that enables efficient identification of dominated subspaces without explicitly computing the dominating space. By reformulating dominated-subspace detection into pairwise dominance tests and decoupling them across independent constraints, DR-BIP eliminates joint optimization and reduces both computational and storage complexity from exponential to polynomial. The resulting dominance components can be precomputed and reused across context changes, further improving efficiency. We formally prove the correctness of DR-BIP and integrate it into several state-of-the-art tree-based complete DCOP algorithms. Extensive empirical evaluation on standard benchmarks confirms that DR-BIP consistently outperforms BIP, significantly reducing both computation time and communication overhead.
Accurate spatio-temporal forecasting requires modeling interactions across multiple spatial and temporal scales. Existing Graph Neural Network (GNN) forecasters primarily operate at a single local scale, limiting their ability to capture global processes that govern system dynamics. Hierarchical GNNs (HGNNs) aim to address this limitation by learning multiscale representations, but in practice, they often fail to preserve global information during coarse-to-fine propagation. We identify this limitation as downward memory loss, where global trends learned at coarse resolutions diminish before influencing fine-grained predictions, leading to misaligned local dynamics. We propose HiGFlow, an HGNN forecaster that explicitly preserves multiscale trends through a self-updating memory buffer that integrates into the coarse-to-fine information flow. This design maintains global contextual signals as an inductive bias throughout hierarchical decoding. We provide a theoretical analysis indicating that when decoding, HiGFlow preserves multiscale trends where residual-based architectures may fail to do so. Empirically, HiGFlow achieves substantially lower MAE and RMSE than state-of-the-art forecasting models across multiple benchmark datasets, demonstrating the importance of explicit global memory in multiscale spatio-temporal forecasting. Our implementation is available at https://github.com/TB862/HiGFlow.
In the field of skeleton-based human action recognition, Graph Convolutional Networks (GCNs) have become a dominant framework. However, existing GCN-based approaches often treat the sequences of two-person interaction as separate entities, ignoring the inherent semantic dependencies and spatial correlations between interacting subjects. Furthermore, high-order skeleton representations naturally exhibit non-Euclidean structures, where Euclidean deep learning models are inherently limited in explicitly capturing and preserving such geometric information. As a countermeasure, we propose a Riemannian Graph Convolutional Network (RGCN) that operates on the Symmetric Positive Definite (SPD) manifolds. Specifically, we model high-order skeletal statistics via Gaussian embedding and propose a Riemannian network to capture inter-subject interactions and global correlations. The proposed RGCN is instantiated under three SPD geometries, and its effectiveness is validated through extensive experiments on three interaction benchmarks. Extensive experimental results show that RGCN provides a competitive and geometrically grounded alternative for skeleton-based interaction recognition.
Recent works in Gaussian Splatting (GS) SLAM highlight the importance of geometric structure. However, existing methods often rely on either 2D or 3D Gaussian primitives, lacking the balance between geometry and appearance, thus failing to precisely model spatial structures. Furthermore, current frameworks either solely utilize SLAM to provide poses for GS or optimize poses and all Gaussian parameters indiscriminately, which neglect geometric consistency constraints essential for robust localization, creating a gap between GS and SLAM. In this paper, a tight-coupling SLAM system based on a Geometric Gaussian Field (GGF) is proposed. First, GGF combines explicit spatial structures with implicit neural residuals to adaptively regulate the Gaussian morphologies, seamlessly transitioning between 2D and 3D Gaussian primitives, ensuring precise spatial modeling. Second, a Geometric Consistency-Guided Refinement (GCGR) strategy is introduced, which exploits depth and normal maps rendered by GGF to construct critical constraints for accurate pose estimation. Simultaneously, updated poses and spatial information are fed back to refine Gaussian morphology, establishing mutual enhancement and tight-coupling between GS and SLAM. Extensive experiments demonstrate that GGF-SLAM outperforms state-of-the-art methods in tracking accuracy, photorealistic rendering, and geometric reconstruction.
The Capacitated Vehicle Routing Problem (CVRP) is a fundamental NP-hard problem with broad applications in logistics and transportation. Real-world CVRPs often involve diverse objectives and complex constraints, such as time windows or backhaul requirements, motivating the development of a unified solution framework. Recent reinforcement learning (RL) approaches have shown promise in combinatorial optimization, yet they rely on end-to-end learning and lack explicit problem-solving knowledge, limiting solution quality. In this paper, we propose a knowledge-embedded framework inspired by the Route-First Cluster-Second heuristics. It incorporates knowledge at two levels: (1) decomposing CVRPs into the route-first and cluster-second subproblems, and (2) leveraging dynamic programming to solve the second subproblem, whose results guide the RL-based constructive solver to solve the first problem. To mitigate partial observability caused by problem decomposition, we introduce a unified history-enhanced context processing module. Extensive experiments show that this framework achieves superior solution quality compared with state-of-the-art learning-based methods, with a smaller gap to classical heuristics, demonstrating strong generalization across diverse CVRP variants.
The rapid advancement of generative AI has made audio deepfakes increasingly indistinguishable from authentic human vocals, posing significant threats to persons-of-interest (POI) such as public figures. Current detection systems primarily rely on generic, black-box models that fail to capture speaker-specific idiosyncratic traits and lack interpretability. In this paper, we propose Phoneme-based Voice Profiling (PVP), a novel personalized defense framework. By shifting the detection paradigm from macro-utterance analysis to micro-phonetic modeling, PVP captures the unique acoustic distributions underlying a POI’s habitual articulatory patterns. Specifically, our framework models speaker-specific phonetic realizations using lightweight Gaussian Mixture Models (GMMs) estimated solely from bona fide reference speech. This design enables data-efficient profiling and robust generalization to previously unseen spoofing attacks without requiring heavy spoof-specific training. Furthermore, we introduce the first large-scale Chinese POI deepfake dataset to benchmark speaker-specific detection. Experimental results demonstrate that PVP significantly outperforms state-of-the-art generic detectors in POI spoofing scenarios, achieving substantial EER reductions while providing fine-grained, phoneme-level interpretability for forensic analysis. Code and data are available at: https://github.com/JunXue-tech/PVP
Link prediction is a foundational task in temporal graphs. While temporal graph neural networks exhibit commendable performance, they are often criticized for providing inadequate representations, especially under limited data. Contrastive learning has been introduced as a solution for graph pre-training to mitigate data scarcity. However, the data augmentation techniques on which they depend frequently lead to suboptimal augmentations, manifesting either as over-augmentation or under-augmentation. In light of these challenges, we explore the under-explored domain of contrastive learning of temporal graph transformers and propose a novel model, ContraTGT, which employs a dual-view graph transformer. This transformer is fine-tuned to extract sequences tailored for specific target nodes, encapsulating both spatial and temporal perspectives. To optimize the contrastive learning of this dual-view transformer, we put forth an innovative, learnable data augmentation method. This technique, which involves selective masking of elements within dual-view sequences, generates superior augmentations, thereby amplifying the potency of the contrastive learning approach. Extensive experiments on five public temporal network datasets demonstrate that our model can consistently outperform all baselines, especially in small training set conditions.
Existing Autonomous Mobility-on-Demand (AMoD) systems and demand-responsive algorithms are designed to adaptively respond to traffic demands, enhancing traffic service quality and profitability. However, such approaches may exacerbate traffic congestion by concentrating vehicles in high-demand areas. In addition, repeated braking, acceleration, and low-speed driving in traffic jams lead to substantial increases in emissions. Moreover, existing Deep Reinforcement Learning (DRL)-based approaches fail to address the instability of cooperative learning in environments with highly variable rewards, such as carbon emissions. To address these issues, we propose the Spatio-Temporal Attention Rebalancing (STAR) framework, leveraging an encoder-decoder architecture. In the proposed framework, we maximize the number of requests served and minimize CO₂ emissions, while ensuring high service quality by limiting maximum travel delays and waiting times for passengers. To evaluate the proposed framework, we conduct simulations in realistic urban traffic scenarios. Experimental results demonstrate that our framework consistently outperforms existing baselines. In a large-scale scenario with a fleet of one hundred vehicles, our approach improves the service rate by up to 10.1% and reduces CO₂ emissions per passenger by up to 69.6% compared to the baseline methods. The proposed framework and baselines are publicly available at https://github.com/2jungeuni/eco-friendly-fleet-rebalancing.
Static benchmarks for LLMs are increasingly compromised by contamination and overfitting, especially on knowledge-intensive reasoning tasks. While recent dynamic benchmarks can alleviate staleness, they often increase difficulty at the expense of answerability and controllability. In this paper, we propose StressEval, a failure-driven data synthesis framework that turns observed model failures into dynamic, challenging, and controlable test instances. StressEval consists of three stages: (i) it constructs a semi-structured difficulty card that identifies the failed reasoning step and its root cause; (ii) it applies a dual-perspective instance-synthesis method that targets both knowledge gaps and reasoning breakdowns while preserving the underlying difficulty factors; and (iii) it applies a gating mechanism to retain only grounded, unambiguous instances. Seeding from multiple knowledge-intensive reasoning datasets, we employ StressEval to build Dynamic-OneEval, a focused suite of challenging dynamic benchmark. Across several state-of-the-art LLMs, Dynamic-OneEval yields substantially larger performance drops than the original benchmarks while retaining explicit difficulty factors, enabling more actionable iteration.
Weakly supervised Referring Expression Comprehension (WREC) aims to localize referred objects using only image-text pairs without box-level annotations. Existing one-stage methods predominantly rely on anchor-level alignment, which suffers from two fundamental limitations: (1) anchors represent local visual patches rather than holistic objects, and (2) they lack the capability to model inter-object relations. To address these issues, we propose OCAAD, an Object-Centric Alignment and Anchor Distillation framework. Our key insight is that different self-attention heads in DINOv2 naturally attend to distinct semantic regions, effectively capturing object-level information. Building on this, OCAAD introduces two complementary modules: (1) Anchor-Object Distillation Module (AODM), which bridges the semantic gap between anchors and objects by transferring object-level knowledge from attention heads to anchors via overlap-aware contrastive learning; and (2) Intra-Modal Relation Consistency (IRC), which explicitly models inter-object relations by enforcing the relational structure among linguistic entities to match that of their visual counterparts. Extensive experiments on RefCOCO, RefCOCO+, and RefCOCOg demonstrate that OCAAD achieves new state-of-the-art performance, validating the effectiveness of objectcentric alignment for WREC. Code is available at https://github.com/VILAN-Lab/OCAAD.
Large Language Models (LLMs) remain highly vulnerable to jailbreak attacks, yet existing evaluations rely primarily on outcome-level metrics such as Attack Success Rate (ASR), providing limited insight into how and why safety failures occur. We propose an explanation-aware safety framework that augments binary harmfulness detection with structured, human-interpretable explanations capturing severity, strategies, trigger spans, rationales, and derived safety factors. To enable scalable and consistent supervision, we introduce a human--LLM hybrid annotation and canonicalization pipeline. We then fine-tune a compact model to generate canonical explanations alongside harmfulness decisions. Across both seen and unseen benchmark settings, our method improves robustness and explanation fidelity. In jailbreak defense evaluation, our approach reduces ASR to 0.44% on Vicuna-7B and 1.30% on GPT-3.5, outperforming existing defense baselines while also achieving the lowest StrongREJECT scores. Beyond outcome-level gains, the model more accurately recovers diagnostic attributes (e.g., attack strategy, trigger spans, and safety factors) than strong general-purpose LLM baselines. Overall, explanation-aware learning exposes diagnostic dimensions that ASR alone cannot capture and provides a more faithful and actionable foundation for robust LLM safety assessment.