arXiv subjects

Zhihao Zhang

Publications and source records attributed to Zhihao Zhang.

At least 19 recordsLinked to original sources

arXiv2026

Collab-Solver: Collaborative Solving Policy Learning for Mixed-Integer Linear Programming

Mixed-integer linear programming (MILP) has been a fundamental problem in combinatorial optimization. Conventional MILP solving mainly relies on carefully designed heuristics embedded in the branch-and-bound framework. Driven by the strong capabilities of neural networks, recent research is exploring the value of machine learning alongside conventional MILP solving. Although learning-based MILP methods have shown great promise, existing works typically learn policies for individual modules in MILP solvers in isolation, without considering their interdependence, which limits both solving efficiency and solution quality. To address this limitation, we propose Collab-Solver, a novel multi-agent-based policy learning framework for MILP that enables collaborative policy optimization for multiple modules. Specifically, we formulate the collaboration between cut selection and branching in MILP solving as a Stackelberg game. Under this formulation, we develop a two-phase learning paradigm to stabilize collaborative policy learning: the first phase performs data-communicated policy pretraining, and the second phase further orchestrates the policy learning for various modules. Extensive experiments on both synthetic and large-scale real-world MILP datasets demonstrate that the jointly learned policies significantly improve solving performance. Moreover, the policies learned by Collab-Solver have also demonstrated excellent generalization abilities across different instance sets.

cs.AI↗

arXiv2026

CAFE: Self-Improving Search Agents Need Co-Evolving Feedback

Reliable search requires more than acquiring external evidence. An agent must also recognize and recover from errors as its trajectory unfolds. In-trajectory feedback provides a mechanism for such recovery by diagnosing where the search has drifted and redirecting subsequent reasoning steps. This is particularly important in long-horizon search, where an early directional error may receive no immediate corrective signal and can compound across later steps. Making such feedback learnable, however, creates a coupled problem: the agent must learn when to request and use feedback, while the critic must learn corrections from outcome-confounded rollouts as the agent's failure patterns evolve. We introduce CAFE (Coupled Agent--Feedback Evolution), a framework in which a shared-parameter model alternates between search-agent and critic roles. CAFE initializes feedback-conditioned recovery from trajectories built around the base agent's own failures, then couples online and offline optimization. During online RL, a comparative feedback estimate uses a prompt-level call--skip success gap to shape request returns, while feedback-aware advantage shaping reweights token advantages before and after feedback. Offline, rollout-derived preference optimization learns feedback from matched successful and unsuccessful trajectories. On seven agentic search benchmarks, CAFE outperforms the evaluated RL-based search agents on average, retains its gains across all six out-of-domain benchmarks, and reduces answer-level hallucinations. One-sided ablations show that improving only the agent or only the critic eventually plateaus, whereas alternating the two updates continues to improve performance. These findings suggest that a self-improving search agent needs feedback that co-evolves with the policy it guides.

cs.AI↗

arXiv2026

ReDraft, Don't Just Distill: Reference-Driven Revision for Continual VLLM Post-Training

Continual post-training of large multimodal models should add new capabilities while preserving those from pre-training, and the two goals pull in opposite directions. SFT gives explicit target supervision that learns a task from near-zero accuracy, but its off-policy targets move the model far enough to cause forgetting; on-policy methods such as RLVR and self-distillation preserve policy proximity yet supply little signal when the policy cannot yet solve the task. We introduce ReDraft (Reference-Driven Revision and Fine-Tuning), which obtains both from the model's own failures: using an expert response only as a reference, it has the model revise its own incorrect rollout, keeps the revision only if a verifier accepts it, and fine-tunes on what survives. Each retained target is therefore explicit, yet still close to the current policy. Across Counting, Clock Reading, and Jigsaw on Qwen2.5-VL-3B/7B, two of them with near zero accuracy, ReDraft gains 56.9 points on the target task against SFT's 52.9 while cutting prior-task loss from 16.6 to 1.5 points (11.3x less forgetting), and improves on OPSD along both axes (19.3 gain, 6.2 loss). Data- and parameter-space analyses match the design: revised targets are more probable under the base model, and the updates they induce stay compact and follow SFT's direction more closely than OPSD's. Repairing the model's own output, rather than replacing it with an expert's, is what lets one objective do both.

cs.AI↗

arXiv2026

MEgoVista: Multi-view Ego-aware Motion Estimation for Metric 4D Hands and Head in the Wild

Learning manipulation from human video requires high-fidelity hand-motion reconstruction in metric units. Today's metric hand labels come from studio rigs and instrumented headsets, and both are confined in the same two ways: neither leaves a prepared setting, and neither is checked against an independent reference. Unconstrained head-worn recording promises the opposite trade-off, scaling with the number of people wearing a device. We therefore introduce MEgoVista, an offline pipeline that turns a single unprepared MEgo View recording into metric two-hand and head motion in one gravity-aligned world frame. Three properties set it apart from existing egocentric reconstruction systems: first, it reconstructs in settings studio volumes and tabletop rigs cannot reach, settling hand ownership at detection so bystander hands stay out of the wearer's trajectory; second, it takes its metric gauge from calibrated stereo rather than a monocular prior, installing scale at initialisation so policies receive physical units, not arbitrary coordinates; third, both outputs are scored inside a motion-capture volume against independent Chingmu optical capture, under a protocol that audits its own reference and charges what a method declines to predict. MEgoVista is offered as a measured route from egocentric video to metric hand supervision, one that widens where such labels can be gathered.

cs.CV↗

arXiv2026

OpWeave: Flexible Operator Disaggregation for Heterogeneous LLM Serving

LLM serving systems increasingly disaggregate inference into finer-grained stages, with recent approaches separating attention from FFN or MoE execution during decode. This operator-level disaggregated serving (ODS) can improve hardware matching and enable independent scaling, particularly across heterogeneous devices. However, existing systems fix operator boundaries and lack a unified characterization of when disaggregation reduces serving cost. We present OpWeave, an end-to-end framework for heterogeneous ODS. OpWeave provides an analytical cost model that bounds the gains of homogeneous and heterogeneous ODS over colocated serving. It jointly optimizes operator partitioning and deployment configuration through a regularity-aware planner that keeps the search tractable even for hybrid-attention models. A vLLM-based runtime executes the synthesized plans with flexible operator stages across heterogeneous device groups. In our evaluation, OpWeave reduces serving cost by up to $1.78\times$ on homogeneous and $1.89\times$ on heterogeneous GPU clusters relative to the best feasible baseline, while meeting latency SLOs.

cs.DC↗

arXiv2026

Neutron Portal and Dark Matter-Baryon Coincidence: from UV Completion to Phenomenology

We present a dynamical solution to the dark matter-baryon coincidence problem based on the neutron portal operator connecting the visible and dark sector asymmetries. This framework is motivated by the possibility that a strongly supercooled dark confinement phase transition accounts for the nano-Hz stochastic gravitational wave signal observed by pulsar timing arrays, while also generating the dark matter and baryon asymmetry in the Universe. We show that the GeV-scale mass of asymmetric dark matter can be naturally correlated with the (multi-)TeV scale cut-off for the neutron portal through its ultraviolet completion. The dark sector is governed by an approximate fixed point and confines once the heavy portal states are integrated out, dynamically generating a scale of $\mathcal{O} ({\rm GeV})$. We analyze both tree and loop-level ultraviolet completions and demonstrate how the resulting confinement scale is linked to the effective neutron portal scale. We also discuss cosmological constraints and experimental prospects in beam dump searches and colliders for probing the neutron portal.

hep-ph↗

arXiv2026

The Minimal Supersymmetric Standard Model with Non-Invertible Selection Rules

We investigate a framework of the Minimal Supersymmetric Standard Model (MSSM) in which the quark and lepton flavor structure and suppression of flavor-changing neutral currents (FCNCs) are governed by non-invertible selection rules. By implementing such non-group-like fusion rules for matter fields, arising from gauging the outer automorphism $\mathbb{Z}_2$ of a discrete $\mathbb{Z}_N$ symmetry, we obtain realistic Yukawa textures that reproduce the observed quark and lepton masses and mixings while ensuring diagonal soft supersymmetry (SUSY) breaking masses and hence suppressing dangerous FCNC processes. We analyze mass insertion parameters under random $\mathcal{O}(1)$ coefficients and find that all flavor-violating effects are consistent with experimental limits on processes such as $μ\to e γ$ and meson mixings. We show that the Yukawa textures and soft terms remain stable under renormalization group evolution. Our results demonstrate that non-invertible selection rules provide a compelling new mechanism to address both the flavor structure and FCNC problems in supersymmetric models.

hep-ph↗

arXiv2026

RefineAny3D: Depth Refinement as Semantic Alignment for Monocular 3D Detection

Monocular 3D object detection spans two regimes: closed-set detectors operating within a fixed category vocabulary, and open-vocabulary detectors that localize arbitrary categories by leveraging depth foundation models for 3D geometry. We find that current depth foundation models, despite their strong zero-shot generalization, lack the object-level precision 3D detection demands: substituting a state-of-the-art depth foundation model for a strong detector's predicted depth degrades accuracy, even falling below the detector's own prediction. Rather than pushing detectors or depth models to be more accurate end-to-end, we treat object-level depth refinement as a stand-alone task and present RefineAny3D, a vision-language model that corrects depth without ever predicting a numerical value. Our key insight is that depth error has a direct visual signature in image space: when projected onto the image, a correctly placed box tightly encloses the object, while a too-far box projects too small and a too-close box projects too large. Depth refinement thus reduces to a visual alignment problem rather than a metric regression problem, which we instantiate by extending the VLM's vocabulary with action tokens that replace numerical depth output with categorical decisions, and by supervising the model on a large-scale chain-of-thought dataset that grounds each decision in explicit visual evidence. Applied as a single post-hoc step, RefineAny3D delivers consistent gains across closed-set detectors, open-vocabulary detectors, and 3D auto-labeling tools, and generalizes to novel categories, scenes, and cameras without retraining.

cs.CV↗

arXiv2026

Evidence-Driven Dynamic Visual Selector for Efficient Long Video Understanding

Recent advancements in MLLM-based long-form video understanding have mitigated inference-time computational cost and limited context lengths by selecting query-relevant frames. However, existing approaches predominantly rely on external proxy scorers and rigid heuristic rules, inevitably suffering from misalignment with the target MLLM's intrinsic evidence and failing to accommodate the non-uniform spatiotemporal information density. In this paper, we propose a fine-grained dynamic visual selection framework named EviSelect, grounded in the target MLLM internal attention evidence. Our method efficiently probes visual evidence via sparse prefilling as a structured prior to guide distribution-aware dynamic sampling. Specifically, we efficiently approximate attention maps of the target MLLM using highly compressed visual inputs and sparse attention, well-aligned to the full counterpart. Conditioned on three complementary attention components derived from this prior, we design a lightweight selector that not only precisely locates query-relevant timestamps but also adaptively adjusts the local sampling rate and spatial resolution. To enable evidence-conditioned spatiotemporal sampling, we formulate the selector as a stochastic policy and optimize it via GRPO under a joint accuracy--efficiency reward. By rewarding correct predictions under lower visual cost through group-relative comparisons, our method encourages the policy to allocate computation dynamically according to the information density of each video. Across three long video understanding benchmarks, EviSelect achieves superior performance compared to existing methods while reducing selected visual tokens by about 50\% and achieving a 3.9x end-to-end speedup.

cs.CV↗

arXiv2026

Fault-tolerant quantum computing with a microwave Cat Bus

The scalability of fault-tolerant neutral-atom quantum computers is constrained by the latency of shuttling with optical tweezers, imposing a stringent trade-off between qubit overhead and circuit depth in quantum algorithm compilation. Here we propose a hardware-efficient, shuttling-free architecture that achieves all-to-all connectivity. Remote Rydberg atoms are resonantly entangled through a microwave ``Cat Bus''---a cavity mode autonomously stabilized in a bosonic cat state. The Cat Bus natively supports the highly parallelized execution of one-to-many $\mathrm{CZ}^n$ gates with exponentially suppressed crosstalk. We derive the resulting cat--atom error channel from the underlying interactions and physical constraints. For fault-tolerant operation, we develop a hardware-aware scheduling scheme that exploits the native cat--atom $\mathrm{CZ}^{n}$ gate to construct a syndrome-extraction circuit with minimum depth. We benchmark the architecture using hypergraph-product (HGP) codes and estimate a 180-fold reduction in syndrome-extraction cycle time at $N=10^5$ data qubits compared with an atom-rearrangement-based architecture. Under matched two-qubit depolarizing noise, the corresponding error threshold increases from $0.55\%$ to $0.72\%$. Under the hardware-derived error model, we obtain a threshold of $0.80\%$, corresponding to a threshold cooperativity of $C_{\mathrm{th}}=7.8 \times 10^4$, compatible with experimentally accessible parameters for Rydberg-coupled microwave-cavity systems. By avoiding atom transport, the Cat Bus provides a route towards high-speed, fault-tolerant neutral-atom quantum computation.

quant-ph↗

arXiv2026

Nonlinear Fourier spectral signatures of rogue waves observed in Bose-Einstein condensates

Modulation instability provides an important framework for understanding rogue wave (RW) formation on continuous backgrounds. However, the formation mechanism and nonlinear spectral structures of RWs in Bose-Einstein condensate (BEC) matter-wave systems with vanishing boundary conditions remain largely unexplored. Here, we employ the nonlinear Fourier transform (NFT), based on the integrable structure of the focusing nonlinear Schrödinger equation and the Zakharov-Shabat scattering problem, to investigate two representative classes of first-order RWs in BEC systems. Through nonlinear spectral analysis and Darboux reconstruction, we demonstrate that both Gaussian-wave-packet-induced extreme localization events and experimentally observed Peregrine solitons are governed by the coherent dynamics of discrete soliton modes encoded in the nonlinear spectrum. For Gaussian initial states, increasing the initial width leads to an increasing number of discrete eigenvalues, resulting in a transition from fundamental solitons and bound states to Christmas-tree-like RW structures. For experimentally observed Peregrine solitons, localized perturbations reshape the discrete spectral configuration and phase evolution, enabling coherent focusing of multiple bound soliton modes. Furthermore, we reveal the spectral mechanism of higher-order RWs and propose an inverse spectral-engineering approach based on discrete-spectrum phase matching. Our results provide a nonlinear spectral perspective for understanding and controlling RW formation in matter-wave systems with vanishing boundary conditions.

nlin.PS↗

arXiv2026

When Agents Go Rogue: Activation-Based Detection of Malicious Behaviors in Multi-Agent Systems

While enabling effective collaboration on complex tasks, LLM-based Multi-Agent Systems (MAS) face critical security challenges due to vulnerabilities at the agent and interaction levels. Most existing MAS security defenses are built upon two core assumptions: semantically-explicit malicious attacks and explicit graph-based modeling of the MAS topology and agent-level interactions. In practice, real-world attacks are becoming more semantically stealthy, while MAS execution is typically asynchronous without the temporal alignment assumed by graph-based propagation models. To address these limitations, we propose AcMAS, an activation-based framework for malicious-behavior detection in MAS. By analyzing internal reasoning states in the activation space of local agents, AcMAS detects even stealthy attacks in a synchronization-robust fashion, without relying on explicit interaction graphs. Moreover, our activation analysis provides critical signals to guide AcMAS in restoring the functionality of compromised agents, rather than the disruptive agent isolation commonly used by the state-of-the-art methods. Comprehensive evaluation demonstrates that AcMAS significantly outperforms graph-based baselines against stealthy attacks, by +0.22 F1 in synchronous settings (0.94 vs. 0.72) and by +0.55 F1 in asynchronous settings (0.93 vs. 0.38), with generalization across diverse open-source LLM backbones, attack intensity, and MAS scale.

cs.CR↗

arXiv2026

AgentGym2: Benchmarking Large Language Model Agents in De-Idealized Real-World Environments

Language agents, i.e., LLM agents, progress rapidly and are increasingly deployed in production environments. This trend underscores the urgent need for rigorous and realistic evaluations. However, most existing benchmarks evaluate agents in simplified, idealized settings. They typically rely on pre-packaged tool interfaces, overlook critical steps, and assume inputs are clean and fully specified. Consequently, they understate the difficulty of real deployments, where uncertainty and noise are ubiquitous and agents must proactively explore the environment to uncover new tools. To bridge this gap, we present AgentGym2, a new evaluation framework with task instances grounded in real-world end-to-end working demands. Beyond reasoning and planning, it measures agents' ability to execute end-to-end procedures, discover tools via exploration, compose tools for unseen tasks, and remain robust to noisy and underspecified information. Experiments on 15 proprietary and open-source models show that even SOTA systems like Gemini and GPT-5 struggle on AgentGym2, revealing a substantial gap between the capability of current agents and the demands of real-world applications.

cs.AI↗

arXiv2026

CORE Planner: Contextual-memory Oriented Reinforcement-learning in Unknown Environments for Robot Navigation

Autonomous navigation in unknown environments requires a robot to efficiently reach a predefined goal while exploring without prior maps. Although progress has been made in this area, most existing works still rely on traditional planning methods with hand-crafted rules, while learning-based methods often suffer from limited environmental memory and challenges in simulation-to-real (sim-to-real) transfer. To overcome these limitations, we propose a Contextual-memory Oriented Reinforcement-learning (CORE) planner for robot navigation in unknown environments. The proposed CORE planner effectively combines the core advantages of traditional and learning-based methods. Specifically, our method uses a sparse visibility graph for structured environment representation, reducing the computational overhead of dense grid maps, and employs a Transformer network to achieve a holistic environmental understanding, thereby significantly improving navigation efficiency. Moreover, we introduce a visibility graph-based graph sparsification method and a contextual memory mechanism, which alleviates local optima and enhances computational performance in large-scale scenes. Finally, our approach achieves zero-shot sim-to-real transfer after training solely on image-based environments, requiring no fine-tuning. Experimental results show that CORE Planner consistently outperforms state-of-the-art methods, including the traditional FAR Planner and all learning-based baselines, across representative environments, reducing travel distance by 13\% over traditional FAR Planner and by up to 48\% relative to learning-based baselines, with larger gains observed in more complex environments. In real-world scenarios, CORE successfully navigates without human intervention, showcasing zero-shot sim-to-real transfer. Code is available at https://github.com/BBD00/core_planner.

cs.RO↗

arXiv2026

Health-ORSC-Bench: A Benchmark for Measuring Over-Refusal and Safety Completion in Health Context

Safety alignment in Large Language Models is critical for healthcare; however, reliance on binary refusal boundaries often results in over-refusal of benign queries or unsafe compliance with harmful ones. While existing benchmarks measure these extremes, they fail to evaluate Safe Completion: the model's ability to maximise helpfulness on dual-use or borderline queries by providing safe, high-level guidance without crossing into actionable harm. We introduce Health-ORSC-Bench, the first large-scale benchmark designed to systematically measure Over-Refusal and Safe Completion quality in healthcare. Comprising 31,920 benign boundary prompts across seven health categories (e.g., self-harm, medical misinformation), our framework uses an automated pipeline with human validation to test models at varying levels of intent ambiguity. We evaluate 30 state-of-the-art LLMs, including GPT-5 and Claude-4, revealing a significant tension: safety-optimised models frequently refuse up to 80% of "Hard" benign prompts, while domain-specific models often sacrifice safety for utility. Our findings demonstrate that model family and size significantly influence calibration: larger frontier models (e.g., GPT-5, Llama-4) exhibit "safety-pessimism" and higher over-refusal than smaller or MoE-based counterparts (e.g., Qwen-3-Next), highlighting that current LLMs struggle to balance refusal and compliance. Health-ORSC-Bench provides a rigorous standard for calibrating the next generation of medical AI assistants toward nuanced, safe, and helpful completions. Furthermore, our benchmark facilitates reproducible evaluation, encourages safety calibration, and supports development of clinically reliable, context-aware, human-aligned medical AI systems. Our code and data are available at: https://github.com/ZhihaoZhang97/Health-ORSC-Bench. Warning: Some contents may include toxic or undesired contents.

cs.AI↗

arXiv2026

pFedUL: Layer-Aware Federated Unlearning for Personalized Federated Learning

Federated unlearning (FU) enables the removal of specific data contributions from federated learning (FL) models to comply with regulations such as the General Data Protection Regulation (GDPR). However, most existing FU methods are designed for the FedAvg paradigm, where all clients share a single global model. In practice, personalized federated learning (pFL) methods such as FedPer, FedRep, Ditto, and FedBN have become widely adopted due to their superior handling of non-IID data. These methods decompose the model into shared global layers and client-specific personalized layers, fundamentally altering the semantics of unlearning, yet this setting has received little attention. We formalize FU under the pFL paradigm, identifying a tension between unlearning completeness on shared layers and personalization preservation for remaining clients. We then propose pFedUL, a layer-aware selective unlearning framework comprising three components: (1) gradient-based layer-wise contribution attribution that separately quantifies the target client's influence on shared and personalized parameters, (2) adaptive selective unlearning that applies differentiated forgetting strategies across layer types, and (3) a lightweight recalibration protocol enabling remaining clients to restore personalization with minimal overhead. We further introduce two new metrics, Personalization Preservation Score (PPS) and Cross-client Fairness Index (CFI), to evaluate pFL-specific unlearning quality. Experiments on CIFAR-10, CIFAR-100, and FEMNIST under varying non-IID settings indicate that pFedUL achieves unlearning effectiveness comparable to full retraining while maintaining an average of 97.3\% personalized accuracy for remaining clients. Compared with six state-of-the-art FU methods adapted to the pFL setting, pFedUL consistently achieves superior personalization preservation.

cs.LG↗

arXiv2026

MPK: A Compiler and Runtime for Mega-Kernelizing Tensor Programs

We introduce Mirage Persistent Kernel (MPK), the first compiler and runtime system that automatically transforms multi-GPU model inference into a single high-performance mega-kernel. MPK introduces an SM-level graph representation that captures data dependencies at the granularity of individual streaming multiprocessors (SMs), enabling cross-operator software pipelining, \rev{fine-grained overlap of computation and communication, and other optimizations that are infeasible under the conventional kernel-per-operator execution model}. The MPK compiler lowers tensor programs into optimized SM-level task graphs and generates fast CUDA implementations for each task, while the MPK in-kernel parallel runtime executes these tasks within a single persistent mega-kernel using decentralized scheduling across SMs. Together, these components provide end-to-end kernel fusion with minimal developer effort, while preserving the flexibility of existing programming models. Our evaluation shows that MPK significantly outperforms existing kernel-per-operator LLM serving systems, achieving up to 1.7$\times$ lower end-to-end inference latency and pushing LLM inference performance close to the limits of the underlying hardware. MPK is publicly available at https://github.com/mirage-project/mirage.

cs.DC↗

arXiv2026

Entropy Polarity in Reinforcement Fine-Tuning: Direction, Asymmetry, and Control

Policy entropy has emerged as a fundamental measure for understanding and controlling exploration in reinforcement learning with verifiable rewards (RLVR) for LLMs. However, existing entropy-aware methods mainly regulate entropy through global objectives, while the token-level mechanism by which sampled policy updates reshape policy entropy remains underexplored. In this work, we develop a theoretical framework of entropy mechanics in RLVR. Our analysis yields a first-order approximation of the entropy change, giving rise to entropy polarity, a signed token-level quantity that predicts how much a sampled update expands or contracts entropy. This analysis further reveals a structural asymmetry: reinforcing frequent high-probability tokens triggers contraction tendencies, whereas expansive tendencies typically require lower-probability samples or stronger distributional correction. Empirically, we show that entropy polarity reliably predicts entropy changes, and that positive and negative polarity branches play complementary roles in preserving exploration while strengthening exploitation. Building on these insights, we propose Polarity-Aware Policy Optimization (PAPO), which preserves both polarity branches and implements entropy control through advantage reweighting. With the empirical entropy trajectory as an online phase signal, PAPO adaptively reallocates optimization pressure between entropy-expanding and entropy-contracting updates. Experiments on mathematical reasoning and agentic benchmarks show that PAPO consistently outperforms competitive baselines, while delivering superior training efficiency and substantial reward improvements.

cs.LG↗