Searcharxiv⌕ Search

arXiv subjects

Deqing Wang

Publications and source records attributed to Deqing Wang.

At least 19 recordsLinked to original sources
arXiv2026

Local-to-Global Sentence-Level Graph Reranking for Scientific Synthesis

Retrieval-augmented scientific synthesis aims to answer complex research questions by integrating information from multiple papers into comprehensive and well-grounded responses. Since the generator can only synthesize the information selected and organized by the reranker, the quality of the generated synthesis depends critically on the reranked results. However, most rerankers operate at the passage level, which leaves key methodological, empirical, and comparative information buried in long and flat contexts, weakening the grounding of generated claims. Moreover, existing rerankers mainly rely on independent query-candidate scoring which overlooks complementary, contextual, and contrasting relations across scientific candidates, limiting information coverage and the comprehensiveness of the resulting synthesis. To address these limitations, we propose LoG-Reranker, a local-to-global sentence-level graph reranking framework for scientific synthesis. LoG-Reranker performs role-aware local scoring to identify fine-grained, query-relevant sentences and then models their relations on a sentence graph across the candidate set to globally refine sentence rankings. Top-ranked sentences and their connected neighbors are organized into a structured input context for generator to produce more grounded and comprehensive synthesis. Extensive experiments on scientific synthesis and reranking benchmarks show that LoG-Reranker consistently outperforms competitive rerankers, yielding more reliable rankings and improving the quality of generated synthesis.

cs.IR↗
arXiv2026

WDL-OPD: Weak-Driven On-Policy Distillation via Mixture-Constrained Co-Training

On-policy distillation (OPD) aligns a student with a teacher on trajectories sampled from the student itself, reducing the train-test state mismatch of offline distillation. The same feedback loop can nevertheless be unstable: each update changes both the policy and the states on which the next update is computed. We introduce WDL-OPD, a mixture-constrained co-training method with two trainable policies. An anchor policy generates every rollout, an auxiliary policy evaluates the same visited states, and a geometric mixture of their token distributions is matched to a frozen teacher by reverse KL. Both policies receive gradient. We show that freezing the auxiliary recovers an anchor-plus-contrast proxy target closely related to OPD$^2$ and W2S-OPD, whereas joint training creates branch-level degrees of freedom that a static delta cannot express. In recorded Qwen3 experiments at 1.7B and 4B scale, WDL-OPD produces the strongest student checkpoint in each of four scale-domain settings. It raises MATH500 accuracy from 0.630 to 0.685 at 4B and from 0.521 to 0.585 at 1.7B. In code generation, seven single-policy OPD configurations exhibit entropy growth or trajectory degradation, while co-training reaches independently re-evaluated development scores of 0.637 and 0.375. Because several comparisons differ in curriculum or initialization, these results support a stabilization hypothesis rather than a universal causal claim. We provide the exact training algorithm, failure evidence, and the controlled comparison matrix needed to test that hypothesis.

cs.LG↗
arXiv2026

Learning to Predict Middle-Layer Attention in MLLMs for Visual Token Pruning

Multimodal large language models (MLLMs) achieve strong performance across diverse vision-language tasks, but their efficiency is limited by the cost of processing numerous visual tokens. Visual token pruning can reduce this cost, but requires accurate token importance estimates. Recent studies have demonstrated that text-to-vision attention from middle language model layers can effectively guide visual token pruning, typically using attention from a predefined middle layer to select the visual tokens to retain. Two problems therefore remain. First, our analysis shows that the layer whose attention is most responsive to the question varies substantially across samples, making a fixed layer suboptimal. Second, obtaining attention from the appropriate middle layer requires processing numerous visual tokens through several language model layers, by which point considerable computation has already been spent. To address both problems, we propose Middle-layer Attention Prediction (MAP), which uses Question Contrastive Teacher Selection to identify a sample-specific teacher layer by contrasting attention under the original and reference questions, and distills attention from the selected layer into a lightweight predictor that estimates visual token importance from multi-modal input features. During inference, MAP combines the predicted importance scores with a diversity criterion to prune visual tokens before the first language model layer. Thus, MAP requires no attention maps for pruning and remains compatible with existing inference acceleration techniques. Across ten benchmarks on LLaVA-NeXT-7B, MAP retains 97.5% of the unpruned model performance with only 5.56% of the visual tokens, yielding a 3.09x end-to-end speedup.

cs.AI↗
arXiv2026

Deferred Exposure of Future Trajectories for Verifiable Reasoning in Autonomous Driving VLMs

Recent Vision-Language-Action (VLA) models for autonomous driving (AD) increasingly utilize chain-of-thought (CoT) supervision to enhance the reasoning capabilities of their Vision-Language Model (VLM) components, yet existing annotation pipelines commonly expose the teacher model to the logged ground-truth (GT) future trajectory. We empirically show that this induces trajectory anchoring bias: teacher models rationalize the revealed outcome rather than infer a decision from scene evidence, producing less causally faithful CoTs and substantially more severe hallucinations, especially in causally challenging scenes. Removing the GT trajectory eliminates this shortcut, but open-ended trajectory generation entangles high-level decision-making with precise geometric synthesis and low-level dynamics. To make trajectory-level driving decisions verifiable without requiring open-ended trajectory synthesis, we introduce Autonomous-Driving Multiple-Choice Question (AD-MCQ), which casts planning as selection among explicit trajectory candidates. Taking this a step further, we propose Deferred Exposure of Future Trajectories for RLVR (DEFT-RLVR) to transform future trajectories from pre-decision anchors into post-decision verification targets. Experimental results show that DEFT-RLVR improves AD reasoning while preserving or even enhancing general visual capabilities. With VLM-only inference and controllable difficulty through candidate construction, AD-MCQ provides a flexible, scalable, and extensible foundation for future research on verifiable AD reasoning.

cs.AI↗
arXiv2026

Achieving Rate-Concurrency Balance for Underwater Concurrent Random Access

Underwater acoustic networks face a fundamental rate--concurrency tradeoff: high-rate waveforms (e.g., OFDM, OTFS) are designed for point-to-point links and rely on orthogonal MAC protocols (e.g., TDMA) to avoid collisions, sacrificing concurrency; conversely, collision-resilient waveforms (e.g., CDMA, ZCMod) support uncoordinated access but are inherently rate-limited by spreading or sparse index modulation. We present \system, a cross-layer concurrent random-access system that combines two new components: (i) \textbf{EZCDM}, an equidistant ZC division-multiplexing waveform that activates multiple cyclic shifts of a ZC root as parallel sub-channels with a tunable rate--robustness tradeoff, and an intra-symbol differential receiver that eliminates the shared multipath channel response without explicit CIR estimation; and (ii) a \textbf{cross-layer link adaptation (LA) framework} featuring beacon-framed random access, user-specific closed-loop power control, and overlap- and CIR-aware common-MS selection. Channel-trace- and signal-trace-driven physical-layer experiments combined with PHY-in-the-loop network simulations demonstrate that \system\ achieves significant BER and throughput gains over conventional waveforms and MAC protocols by converting traditionally destructive collisions into decodable concurrent streams.

cs.NI↗
arXiv2026

Multi-Objective Exploration and Preference Optimization via Mutual Information

Aligning large language models with diverse and heterogeneous human values requires multi-objective alignment methods to effectively trade off conflicting preference dimensions. Current methods achieve this trade-off by training policies conditioned on preference vectors and leveraging online direct preference optimization. However, exploration uncertainty can cause the reward distributions of responses generated under different preference vectors to overlap, and the generated responses may fail to effectively align with the corresponding preference vectors. In this paper, we propose Multi-Objective Exploration and Preference Optimization via Mutual Information (MI-EPO), an information-theoretic framework. It unifies multi-objective exploration and alignment by maximizing the joint conditional mutual information among generated responses, preference feedback, and preference vectors. By incorporating a probabilistic routing mechanism, MI-EPO naturally decomposes objective alignment and preference-aware exploration, encouraging the model to generate responses that are distinguishable and aligned with different preference conditions. Experiments on safe alignment and helpful assistant tasks show that MI-EPO significantly improves the alignment between generated responses and preference vectors, makes the outputs more controllable, and achieves stable trade-offs across multiple objectives.

cs.CL↗
arXiv2026

SynGR: Unleashing the Potential of Cross-Modal Synergy for Generative Recommendation

Generative Recommendation (GR) has emerged as a promising paradigm by formulating item recommendation as a sequence-to-sequence generation task over item identifiers. Recent studies have incorporated multimodal signals to provide richer token-level evidence for generation. However, existing approaches largely rely on alignment-centric fusion and underexplore synergistic information across modalities. In practice, synergistic information plays a critical role in capturing emergent item properties that cannot be inferred from any single modality alone. Such properties encode intrinsic item semantics and guide user preferences, enabling models to move beyond surface-level feature matching. To address this limitation, we propose \textbf{SynGR}, a synergistic generative recommendation framework that explicitly encourages the exploitation of cross-modal dependencies during generation. By constraining overreliance on dominant modalities, SynGR enables the model to capture emergent item semantics beyond shared or modality-specific signals. Extensive experiments across three benchmark datasets demonstrate that SynGR achieves superior performance.

cs.IR↗
arXiv2026

Route Before Retrieve: Activating Latent Routing Abilities of LLMs for RAG vs. Long-Context Selection

Recent advances in large language models (LLMs) have expanded the context window to beyond 128K tokens, enabling long-document understanding and multi-source reasoning. A key challenge, however, lies in choosing between retrieval-augmented generation (RAG) and long-context (LC) strategies: RAG is efficient but constrained by retrieval quality, while LC supports global reasoning at higher cost and with position sensitivity. Existing methods such as Self-Route adopt failure-driven fallback from RAG to LC, but remain passive, inefficient, and hard to interpret. We propose Pre-Route, a proactive routing framework that performs structured reasoning before answering. Using lightweight metadata (e.g., document type, length, initial snippet), Pre-Route enables task analysis, coverage estimation, and information-need prediction, producing explainable and cost-efficient routing decisions. Our study shows three key findings: (i) LLMs possess latent routing ability that can be reliably elicited with guidelines, allowing single-sample performance to approach that of multi-sample (Best-of-N) results; (ii) linear probes reveal that structured prompts sharpen the separability of the "optimal routing dimension" in representation space; and (iii) distillation transfers this reasoning structure to smaller models for lightweight deployment. Experiments on LaRA (in-domain) and LongBench-v2 (OOD) confirm that Pre-Route outperforms Always-RAG, Always-LC, and Self-Route baselines, achieving superior overall cost-effectiveness.

cs.CL↗
arXiv2026

Iterative Receiver Processing at Relays in PNC-Enabled Multi-Hop Underwater Acoustic Networks

Physical-layer network coding (PNC) can increase end-to-end throughput in bi-directional multi-hop underwater acoustic (UWA) networks. However, multipath delay spread and Doppler-induced inter-carrier interference (ICI) in UWA channels can degrade the reliability of PNC transmission in a three-node relay configuration. More critically, error accumulation across multiple relay nodes leads to a pronounced increase in the end-to-end bit error rate (BER) in multi-hop networks. To address this issue, we develop an iterative detection and decoding processing strategy for relay nodes within a PNC-enabled multi-hop UWA network based on orthogonal frequency division multiplexing (OFDM) modulation. The proposed design integrates three key algorithms: (i) an adaptive channel-aware factor graph detection algorithm that is suited for time-varying UWA channels; (ii) a parity-check-constrained soft-information refinement algorithm that improves the accuracy of the information feedback from the decoder to the detector; and (iii) a linear minimum mean square error (LMMSE) detection algorithm based on a superimposed model, which offers low computational complexity as an alternative scheme. Extensive simulation results demonstrate that the adaptive detection algorithm achieves BERs on the order of $10^{-5}$ at a relative velocity of 1.5 m/s UWA channel and a signal-to-noise (SNR) of 8~dB. Both lake experiments and sea trials in the Taiwan Strait confirm that the proposed iterative receiver algorithms outperform baseline schemes in terms of BER performance under practical UWA channel conditions, showing their robustness and applicability in real multi-hop deployments.

cs.NI↗
arXiv2026

Look One Step Ahead: Forward-Looking Incentive Design with Strategic Privacy for Proactive Service Provisioning over Air-Ground Integrated Edge Networks

In air-ground integrated networks (AGINs), unmanned aerial vehicles (UAVs) provide on-demand edge services to ground vehicles. Realizing this vision requires carefully designed incentives to coordinate interactions among self-interested participants. This is exacerbated by the dynamic nature of AGINs, where spatio-temporal variations introduce significant uncertainty in matching UAVs and vehicles. Existing real-time service provisioning typically relies on precise trajectory information, raising privacy concerns and incurring decision latency. To address these challenges, we propose look one-step ahead (LOSA), a novel framework for efficient and privacy-aware service provisioning. By exploiting predictable vehicle travel times between intersections, LOSA decomposes the process into two coupled phases: (i) a privacy-aware look-ahead phase and (ii) a lightweight real-time execution phase. The look-ahead phase allows vehicles to adaptively adjust privacy budgets based on historical utility, balancing trajectory exposure and matching accuracy. Leveraging this, a double auction mechanism establishes binding one-step-ahead agreements (OSAAs) through trajectory similarity clustering, while constructing preference lists to hedge against mobility uncertainty. The execution phase then enforces pre-established OSAAs and preference lists, resolving real-time resource conflicts without costly re-negotiations. This design reduces computational overhead and preserves robustness. We analytically corroborate that LOSA guarantees truthfulness, individual rationality, and budget balance. Experiments on real-world datasets (DAIR-V2X, HighD, and RCooper) demonstrate that LOSA achieves superior privacy protection while lowering transaction latency compared to baseline approaches.

cs.NI↗
arXiv2026

Policy Improvement Reinforcement Learning

Reinforcement learning has become a central post-training paradigm for improving LLM and agent capabilities. Yet existing RL post-training methods share a common blind spot: they construct local learning signals from sampled trajectories, rewards, or feedback-conditioned targets, then update the policy without explicitly verifying whether the resulting policy outperforms its predecessor. Optimizing these local signals does not necessarily produce a better policy, while finite sampling, generation stochasticity and feedback noise can further widen this gap. We argue that the missing ingredient is policy improvement feedback: the ability to measure progress across policy iterations. We introduce Policy Improvement Reinforcement Learning (PIRL), which formulates inter-iteration performance gain as an explicit objective structurally aligned with final task performance. Building on PIRL, we propose Policy Improvement Policy Optimization (PIPO), a plug-in closed-loop framework that verifies the previous update against a sliding-window historical performance anchor. PIPO uses this improvement feedback to modulate the local learning signal of the base policy optimization algorithm, reinforcing updates associated with measured progress and suppressing those associated with performance drops. We provide theoretical evidence that PIPO locally aligns policy updates with the PIRL improvement objective. Experiments on mathematical reasoning, code, tool-use, and self-distillation settings show that PIPO yields consistent gains across PPO, group-relative, and self-distillation policy optimization families.

cs.LG↗
arXiv2026

Heterogeneous Agent Collaborative Reinforcement Learning

We introduce Heterogeneous Agent Collaborative Reinforcement Learning (HACRL), a new Reinforcement Learning from Verifiable Reward (RLVR) problem that addresses the inefficiencies of isolated multi-agent on-policy optimization. HACRL enables collaborative optimization with independent execution: heterogeneous agents share verified rollouts during training to mutually improve, while operating independently at inference time. Unlike LLM-based multi-agent reinforcement learning (MARL), HACRL does not require coordinated deployment, and unlike on-/off-policy distillation, it enables bidirectional mutual learning among heterogeneous agents rather than one-directional homogeneous teacher-to-student transfer. Building on this problem, we propose HACPO, a collaborative RL algorithm that enables principled rollout sharing to maximize sample utilization and cross-agent knowledge transfer. To mitigate capability discrepancies and policy distribution shifts, HACPO introduces four tailored mechanisms with theoretical guarantees on unbiased advantage estimation. Extensive experiments across diverse heterogeneous model combinations and reasoning benchmarks show that HACPO consistently improves all participating agents, outperforming GSPO with double rollouts by an average of 3.6% while using only half the rollout cost.

cs.LG↗
arXiv2026

UniFAR: A Unified Facet-Aware Retrieval Framework for Scientific Documents

Scientific document retrieval (SDR) plays a critical role in modern scientific research, supporting knowledge discovery and evidence-based reasoning. It has evolved along two paradigms: document--document (doc-doc) retrieval driven by inter-document contrastive learning, and question--document (q-doc) retrieval emerging from LLMs and RAG for natural-language interaction. In practice, scientific workflows rely on both paradigms, requiring retrieval of related papers given a seed document and identifying relevant documents given a user question. However, existing methods typically treat these paradigms separately, hindering their complementary strengths. To address this, we propose UniFAR, a unified facet-aware retrieval framework that jointly supports doc-doc and q-doc retrieval within a shared representation space. UniFAR introduces a multi-granularity representation and aggregation module to unify the encoding of short questions and long documents, and a facet-level modeling mechanism with learnable anchors to capture structured semantic roles and complex user intents. It further adopts a facet-aware joint training strategy that integrates doc-doc and q-doc contrastive objectives with facet-level alignment, enabling unified learning from both inter-document relations and question-oriented supervision. Experiments on three benchmark datasets under both doc-doc and q-doc settings across multiple backbone models show that UniFAR consistently outperforms strong baselines and generalizes effectively across different backbones.

cs.IR↗
arXiv2026

UniARM: Towards a Unified Autoregressive Reward Model for Multi-Objective Test-Time Alignment

Multi-objective alignment aims to align LLM responses with multiple human preference objectives. Among existing methods, guiding the generation of frozen LLMs through autoregressive reward models (ARMs) to accomplish multi-objective test-time alignment is a low-cost solution. However, these methods typically rely on independent parameters for each preference objective, either by training ARMs independently across preference dimensions, which neglects interactions among preference features, or by training a single ARM with separate feature extraction modules for each preference, which can cause feature entanglement. Both strategies can result in misalignment between generated outputs and user preferences. To address this limitation, we propose Preference-Modulated \& Shared Low-Rank Adaptation (MoSLoRA) for ARM training, which first extracts shared features via a preference-agnostic module and then applies affine transformations to shared features via a preference modulation module conditioned on mixed preference vectors. This design mitigates feature entanglement and enables precise control over preference trade-offs during inference. Building on this, we introduce the Unified Autoregressive Reward Model (UniARM), a novel framework for multi-objective test-time alignment. UniARM jointly models all preference dimensions in a single parameter space, eliminating the need for independent parameters for each preference objective. es on larger-scale LLMs, enhancing its practical usability.

cs.CL↗
arXiv2026

Weak-Driven Learning: How Weak Agents make Strong Agents Stronger

As post-training optimization becomes central to improving large language models, we observe a persistent saturation bottleneck: once models grow highly confident, further training yields diminishing returns. While existing methods continue to reinforce target predictions, we find that informative supervision signals remain latent in models' own historical weak states. Motivated by this observation, we propose WMSS (Weak Agents Can Make Strong Agents Stronger), a post-training paradigm that leverages weak checkpoints to guide continued optimization. By identifying recoverable learning gaps via entropy dynamics and reinforcing them through compensatory learning, WMSS enables strong agents to improve beyond conventional post-training saturation. Experiments on mathematical reasoning and code generation datasets show that agents trained with our approach achieve effective performance improvements, while incurring zero additional inference cost.

cs.AI↗
arXiv2026

Does Your Reasoning Model Implicitly Know When to Stop Thinking?

Recent advancements in large reasoning models (LRMs) have greatly improved their capabilities on complex reasoning tasks through Long Chains of Thought (CoTs). However, this approach often results in substantial redundancy, impairing computational efficiency and causing significant delays in real-time applications. Recent studies show that longer reasoning chains are frequently uncorrelated with correctness and can even be detrimental to accuracy. In a further in-depth analysis of this phenomenon, we surprisingly uncover and empirically verify that LRMs implicitly know the appropriate time to stop thinking, while this capability is obscured by current sampling paradigms. Motivated by this, we introduce SAGE (Self-Aware Guided Efficient Reasoning), a novel sampling paradigm that unleashes this efficient reasoning potential. Furthermore, integrating SAGE as mixed sampling into group-based reinforcement learning (SAGE-RL) enables SAGE-RL to effectively incorporate SAGE-discovered efficient reasoning patterns into standard pass@1 inference, markedly enhancing both the reasoning accuracy and efficiency of LRMs across multiple challenging mathematical benchmarks.

cs.AI↗
arXiv2026

Contextual Rollout Bandits for Reinforcement Learning with Verifiable Rewards

Reinforcement Learning with Verifiable Rewards (RLVR) is an effective paradigm for improving the reasoning capabilities of large language models. However, existing RLVR methods utilize rollouts in an indiscriminate and short-horizon manner: responses of heterogeneous quality within each prompt are treated uniformly, and historical rollouts are discarded after a single use. This leads to noisy supervision, poor sample efficiency, and suboptimal policy updates. We address these issues by formulating rollout scheduling in RLVR as a contextual bandit problem and proposing a unified neural scheduling framework that adaptively selects high-value rollouts throughout training. Each rollout is treated as an arm whose reward is defined by the induced performance gain between consecutive optimization steps. The resulting scheduler supports both noise-aware intra-group selection and adaptive global reuse of historical rollouts within a single principled framework. We provide theoretical justification by deriving sublinear regret bounds and showing that enlarging the rollout buffer improves the achievable performance upper bound. Experiments on six mathematical reasoning benchmarks demonstrate consistent gains in performance and training efficiency across multiple RLVR optimization methods.

cs.LG↗
arXiv2026

Real-Time Aligned Reward Model beyond Semantics

Reinforcement Learning from Human Feedback (RLHF) is a pivotal technique for aligning large language models (LLMs) with human preferences, yet it is susceptible to reward overoptimization, in which policy models overfit to the reward model, exploit spurious reward patterns instead of faithfully capturing human intent. Prior mitigations primarily relies on surface semantic information and fails to efficiently address the misalignment between the reward model (RM) and the policy model caused by continuous policy distribution shifts. This inevitably leads to an increasing reward discrepancy, exacerbating reward overoptimization. To address these limitations, we introduce R2M (Real-Time Aligned Reward Model), a novel lightweight RLHF framework. R2M goes beyond vanilla reward models that solely depend on the semantic representations of a pretrained LLM. Instead, it leverages the evolving hidden states of the policy (namely policy feedback) to align with the real-time distribution shift of the policy during the RL process. This work points to a promising new direction for improving the performance of reward models through real-time utilization of feedback from policy models.

cs.AI↗