Search arXiv⌕ Search

arXiv subjects

Hao Zhou

Publications and source records attributed to Hao Zhou.

At least 109 records · Page 6Linked to original sources
arXiv2026

NaviAgent: Graph-Driven Bilevel Planning for Scalable Tool Orchestration

Large Language Models (LLMs) increasingly act as function-call agents that invoke external tools to tackle tasks beyond their static knowledge. However, they typically invoke tools one at a time without a global view of task structure. As tools often depend on one another, this leads to error accumulation and poor scalability, particularly when scaling to hundreds or thousands of tools. To address these limitations, we propose NaviAgent, an explicit bilevel architecture that decouples task planning from tool execution through graph-based modeling of tool relations. At the planning level, the LLM-based agent decides whether to respond directly, clarify intent, or retrieve and execute a toolchain independent of inter-tool complexity. At the execution level, a Tool World Navigation Model (TWNM) encodes structural and behavioral relations among tools, steering the agent to compose scalable and robust invocation sequences. Incorporating feedback from real tool interactions, NaviAgent achieves closed-loop alignment between planning and execution, enabling adaptive navigation in large-scale tool ecosystems. Evaluations on API-Bank and ToolBench show consistent improvements in task success rate (TSR), with TWNM yielding an average gain of 13.1 points on complex tasks. Further tests on 50 real APIs across 7 domains show consistent gains of 4.3--12.0 points, with fewer steps and latency, demonstrating robust generalization under real-world dynamics.

cs.AI↗
arXiv2026

Investigating Cross-Modal Skill Injection: Scenarios, Methods, and Hyperparameters

Vision-Language Models (VLMs) have demonstrated remarkable proficiency in general multi-modal understanding; yet they struggle to efficiently acquire continually evolving domain-specific skills. Conventional approaches to enhancing VLM capabilities, such as Supervised Fine-Tuning (SFT), require extensive dataset curation and substantial computational resources. Model merging has emerged as an efficient alternative that enables the transfer of domain-specific expertise from Large Language Models (LLMs) to VLMs without incurring additional training data requirements or significant computational overhead. Unlike conventional merging of homogeneous LLMs, which mainly aggregates existing capabilities, cross-modal skill injection aims to induce emergent cross-modal capabilities by integrating a domain-expert LLM into a VLM. However, existing research lacks a systematic analysis of the applicability and methodology of cross-modal skill injection. In this study, we investigate cross-modal skill injection across three main aspects: scenarios, methods, and hyperparameters. For scenarios, we find that cross-modal skill injection generally performs well in instruction-following and cross-lingual settings, yet struggles with mathematical reasoning. For methods, we find that classic approaches such as TA and DARE consistently achieve superior performance over alternative merging methods. We also provide a systematic and quantitative analysis of the hyperparameter tuning that these classic methods critically depend on.

cs.CL↗
arXiv2026

A Closed-loop, State-centric, Multi-agent Framework for Passenger Load Estimation from Heterogeneous Data Streams

To support operations and passenger-facing services, transit agencies need reliable passenger load trajectories. Currently, load estimates are typically inferred from imperfect sensing systems rather than fully observed, and the accuracy of modern automatic passenger counting (APC) systems still varies with station layout, flow intensity, and operating conditions. To address the challenges of robust passenger load estimation from heterogeneous data streams, including incremental count errors, evidence conflicts, and context-dependent sensor reliability, we propose a closed-loop, state-centric, multi-agent framework. This method enforces physical feasibility at every step, allocates trust dynamically among evidence sources, and feeds physics-derived violation residuals back into training for robustness improvement. The architecture consists of a unified stop-event backbone, a coupled Perception--Physical--Fusion loop for stop-by-stop inference, and optional trip-level macro-correction and closed-loop calibration modules.

cs.LG↗
arXiv2026

PEGRL: Improving Machine Translation by Post-Editing Guided Reinforcement Learning

Reinforcement learning (RL) has shown strong promise for LLM-based machine translation, with recent methods such as GRPO demonstrating notable gains; nevertheless, translation-oriented RL remains challenged by noisy learning signals arising from Monte Carlo return estimation, as well as a large trajectory space that favors global exploration over fine-grained local optimization. We introduce \textbf{PEGRL}, a \textit{two-stage} RL framework that uses post-editing as an auxiliary task to stabilize training and guide overall optimization. At each iteration, translation outputs are sampled to construct post-editing inputs, allowing return estimation in the post-editing stage to benefit from conditioning on the current translation behavior, while jointly supporting both global exploration and fine-grained local optimization. A task-specific weighting scheme further balances the contributions of translation and post-editing objectives, yielding a biased yet more sample-efficient estimator. Experiments on English$\to$Finnish, English$\to$Turkish, and English$\leftrightarrow$Chinese show consistent gains over RL baselines, and for English$\to$Turkish, performance on COMET-KIWI is comparable to advanced LLM-based systems (DeepSeek-V3.2). Our code and a set of representative pretrained models are publicly available at \url{https://github.com/NJUNLP/peg-rl} and \url{https://huggingface.co/collections/DGME/pegrl}

cs.CL↗
arXiv2026

A Data-Efficient Path to Multilingual LLMs: Language Expansion via Post-training PARAM$Δ$ Integration into Upcycled MoE

Expanding Large Language Models~(LLMs) to new languages is a costly endeavor, demanding extensive Continued Pre-Training~(CPT) and data-intensive alignment. While recent data-free merging techniques attempt to bypass alignment by fusing a multilingual CPT-enhanced model with its instruct counterpart, they are plagued by a critical trade-off: mitigating parameter conflicts to preserve original abilities inevitably dilutes new language acquisition, and vice-versa. To resolve this conflict, we introduce \method, which upcycles a dense model into a Mixture-of-Experts~(MoE) architecture, allocating different experts to different languages. Alignment ability is then transferred by grafting a MoE-expanded parameter delta~($Δ_{\text{post}}$) to the CPT-enhanced base model, bypassing the complex alignment phase. Experiments demonstrate \method's superiority even against baselines with similar FLOPs or number of parameters; it improves performance on expanded languages while effectively preserving original capabilities. We further show our approach is highly applicable across different models and Post-training deltas.

cs.CL↗
arXiv2026

PerFlow: Physics-Embedded Rectified Flow for Efficient Reconstruction and Uncertainty Quantification of Spatiotemporal Dynamics

Reconstructing PDE-governed fields from sparse and irregular measurements is challenging due to their ill-posed nature. Deterministic surrogates are trained on dense fields that struggle with limited measurements and uncertainty quantification. Generative models, by learning distributions over spatiotemporal fields, can better handle sparsity and uncertainty. However, existing generative approaches enforce data consistency and PDE constraints simultaneously via sampling-time gradient guidance, resulting in slow and unstable inference. To this end, we propose PerFlow, a Physics-embedded rectified Flow for efficient sparse reconstruction and uncertainty quantification of spatiotemporal dynamics. PerFlow decouples observation conditioning from physics enforcement, performing guidance-free conditioning by feeding observations into rectified-flow dynamics while embedding hard physics via a constraint-preserving projection (e.g., incompressibility or conservation). Theoretically, we establish invariance guarantees to ensure that trajectories remain on the physics-consistent manifold throughout sampling. Experiments on various PDE systems demonstrate competitive reconstruction accuracy with sound physics consistency, while enabling efficient conditional sampling (e.g., 50 steps) and up to 320x faster inference than 2000-step guided diffusion baselines.

cs.LG↗
arXiv2026

Hierarchical Attacks for Multi-Modal Multi-Agent Reasoning

Multi-modal multi-agent systems (MM-MAS) have gained increasing attention for their capacity to enable complex reasoning and coordination across diverse modalities. As these systems continue to expand in scale and functionality, investigating their potential vulnerabilities has become increasingly important. However, existing studies on adversarial attacks in multi-agent systems primarily focus on isolated agents or unimodal settings, leaving the vulnerabilities of MM-MAS largely underexplored. To bridge this gap, we introduce HAM$^{3}$, a Hierarchical Attack framework for multi-modal multi-agent systems that decomposes attacks into three interconnected layers. Specifically, at the perception layer, HAM$^{3}$ mounts attacks by perturbing visual inputs, textual inputs, and their fused visual-textual representations. At the communication layer, it performs communication-level attacks that corrupt message content and interaction topology, such as manipulating shared context or communication links to distort collective information flow. At the reasoning layer, it conducts reasoning-level attacks that interfere with each agent's cognitive pipeline, biasing reasoning trajectories and ultimately compromising final decisions. We evaluate HAM$^{3}$ on the GQA benchmark through multi-agent systems built on distinct reasoning paradigms including ReAct, Plan-and-Solve, and Reflexion. Experiments demonstrate that our framework achieves an Attack Success Rate of up to 78.3%, with reasoning-layer attacks being the most effective. More than half of the successful attacks lead multiple agents to produce consistent errors. These findings offer valuable insights for building more robust and interpretable multi-agent intelligence.

cs.AI↗
arXiv2026

Hierarchical LLM-Driven Control for HAPS-Assisted UAV Networks: Joint Optimization of Flight and Connectivity

Uncrewed aerial vehicles (UAVs) are increasingly deployed in complex networked environments, yet the joint optimization of multi-UAV motion control and connectivity remains a fundamental challenge. In this paper, we study a multi-UAV system operating in an integrated terrestrial and non-terrestrial network (ITNTN) comprising terrestrial base stations and high-altitude platform stations (HAPS). We consider a three-dimensional (3D) aerial highway scenario where UAVs must adapt their motion to ensure collision avoidance, efficient traffic flow, and reliable communication under dynamic and partially observable conditions. We first model the problem as a hierarchical multi-objective partially observable Markov decision process (H-MO-POMDP), capturing the coupling between control and communication objectives. Based on this formulation, we propose a large language model (LLM)-driven hierarchical multi-rate control framework. At the global level, an LLM-based controller on the HAPS performs long-term planning for load balancing and handover decisions. At the local level, each UAV employs a hybrid controller that integrates a slow-timescale LLM for high-level spatial reasoning with a reinforcement learning agent for faster UAV-to-infrastructure (U2I) communication and motion control. We further develop a high-fidelity 3D simulation platform by integrating the gym-pybullet-drones environment with 3GPP-compliant RF/THz channel models. Numerical results demonstrate that the proposed framework significantly outperforms state-of-the-art baselines, achieving a 14% increase in transportation efficiency and a 25% improvement in telecommunication throughput. Additionally, it achieves a 23% reduction in physical collision rates, demonstrating strong handover stability and zero-shot generalization in dynamic scenarios.

cs.AI↗
arXiv2026

Safety Anchor: Defending Harmful Fine-tuning via Geometric Bottlenecks

The safety alignment of Large Language Models (LLMs) remains vulnerable to Harmful Fine-tuning (HFT). While existing defenses impose constraints on parameters, gradients, or internal representations, we observe that they can be effectively circumvented under persistent HFT. Our analysis traces this failure to the inherent redundancy of the high-dimensional parameter space: attackers exploit optimization trajectories that are orthogonal to defense constraints to restore harmful capabilities while deceptively adhering to safety restrictions. To address this, we propose Safety Bottleneck Regularization (SBR). SBR shifts the defensive focus from the redundant parameter space to the unembedding layer, which serves as a geometric bottleneck. By anchoring the final hidden states of harmful queries to those of the safety-aligned model, SBR enables the model to maintain safe responses even under persistent HFT. Extensive experiments confirm SBR's effectiveness, demonstrating that utilizing just a single safety anchor is sufficient to reduce the Harmful Score to $<$10 while preserving competitive performance on benign downstream tasks.

cs.CR↗
arXiv2026

The Double-Burst Nature and Early Afterglow Evolution of Long GRB 110801A

We present a comprehensive temporal and spectral analysis of the long-duration gamma-ray burst GRB 110801A, utilizing multi-band data from the Neil Gehrels Swift Observatory and ground-based telescopes. The $γ$-ray emission exhibits a distinct two-episode (``double-burst'') structure. Rapid follow-up observations in the optical and X-ray bands provide full coverage of the second burst. The optical light curve begins to rise approximately 135 s after the trigger, significantly preceding the second emission episode observed in X-rays and $γ$-rays at $\sim 320$ s. This chromatic behavior suggests different physical origins for the optical and high-energy emissions. Joint broadband spectral fitting (optical to $γ$-rays) during the second episode reveals that a two-component model, consisting of a power-law plus a Band function, provides a superior fit compared to single-component models. We interpret the power-law component as the afterglow of the first burst (dominating the optical band), while the Band component is attributed to the prompt emission of the second burst (dominating the high-energy bands). A physical synchrotron model is also found to be a viable candidate to explain the high-energy emission. Regarding the afterglow, the early optical light curve displays a sharp transition from a rise of $\sim t^{2.5}$ to $\sim t^{6.5}$, which is well-explained by a scenario involving both reverse shock (RS) and forward shock (FS) components. We constrain the key physical parameters of the burst, deriving an initial Lorentz factor $Γ_0 \sim 60$, a jet half-opening angle $θ_j \sim 0.09$, and an isotropic kinetic energy $E_{\rm k,iso} \sim 10^{54.8}$ erg.

astro-ph.HE↗
arXiv2026

Cut-In Gap Acceptance Toward Autonomous vs. Human-Driven Vehicles: Evidence from the Waymo Open Motion Dataset

Autonomous vehicles (AVs) are widely known to follow conservative, rule-based motion policies that surrounding drivers can learn to anticipate. A direct consequence is that human drivers may accept shorter longitudinal gaps when cutting in front of an AV than when targeting another human-driven vehicle (HDV). We test this hypothesis using the Waymo Open Motion Dataset (WOMD), which provides 25,906 real-world highway scenarios at 10 hertz. An eight-criterion lane-change detector extracts 706 HDV-to-AV and 3,172 HDV-to-HDV cut-in events from the same traffic environment. The median accepted gap in front of the Waymo AV is 7.58 meters versus 9.57 meters for HDV targets, a 1.99 meter reduction that is statistically significant (p equals 5.76 times 10 to the negative eighth power, d equals negative 0.224) and persists under speed-matched resampling. Cut-in speeds toward the AV are 37 percent higher (51.7 versus 37.7 kilometers per hour, d equals 0.502), and 68.0 percent of AV-targeted cut-ins occur below the 10 meter gap boundary versus 51.8 percent of HDV-targeted events (chi-squared equals 60.5, p is less than 10 to the negative thirteenth power). These results reveal a systematic and safety-relevant asymmetry in human gap-acceptance behavior that warrants AV-specific calibration of both motion-planning safety envelopes and traffic simulation models.

cs.RO↗
arXiv2026

DataArc-SynData-Toolkit: A Unified Closed-Loop Framework for Multi-Path, Multimodal, and Multilingual Data Synthesis

Synthetic data has emerged as a crucial solution to the data scarcity bottleneck in large language models (LLMs), particularly for specialized domains and low-resource languages. However, the broader adoption of existing synthetic data tools is severely hindered by convoluted workflows, fragmented data standards, and limited scalability across modalities. To address these limitations, we develop DataArc-SynData-Toolkit, an open-source framework featuring: (1) a configuration-driven, end-to-end pipeline equipped with an intuitive visual interface and simplified CLI for exceptional usability; (2) a unified, quality-controllable synthesis paradigm that standardizes multi-source data generation to ensure high reusability; and (3) a highly modular architecture designed for seamless multimodal, multilingual, and multi-task adaptation. We apply the toolkit in multiple application scenarios. Experimental results demonstrate that our toolkit achieves an optimal balance between generation efficiency and data quality. By offering an end-to-end and visually interactive pipeline, DataArc-SynData-Toolkit significantly lowers the technical barrier to synthetic data generation and subsequent model training, accelerating its practical deployment in real-world applications.

cs.LG↗
arXiv2026

Physiology-Aware Masked Cross-Modal Reconstruction for Biosignal Representation Learning

Biosignals acquired from different locations on the body often provide temporally ordered views of the same underlying physiological process. However, most existing self supervised learning methods treat these signals as interchangeable views, overlooking the directional temporal dynamics that link them. A canonical example is the relationship between electrocardiography (ECG), which captures the electrical activation initiating each heartbeat, and photoplethysmography (PPG), which records the resulting peripheral pulse delayed by vascular dynamics. To capture this structured relationship, we introduce xMAE, a biosignal pretraining framework that leverages masked cross modal reconstruction across temporally ordered biosignals as a training time constraint to encourage physiologically meaningful timing structure in the learned representations. We show that pretraining with xMAE yields representations that outperform both unimodal and multimodal baselines on 15 of 19 downstream tasks, including cardiovascular outcome prediction, abnormal laboratory test detection, sleep staging, and demographic inference, while generalizing across devices, body locations, and acquisition settings. Further analysis suggests that the ECG PPG timing structure is reflected in the learned PPG representations. More broadly, xMAE demonstrates the effectiveness of incorporating temporal structure into multimodal pretraining when signals observe different stages of a shared underlying process. Code is available at https://github.com/hzhou3/xMAE.

cs.LG↗
arXiv2026

Design and Evaluation of a PMT High-Voltage system for Deepsea Neutrino Telescope

We present the design and characterization of a Cockcroft--Walton (CW) high-voltage (HV) system developed for deep-sea neutrino telescopes. The system provides independently adjustable bias voltages for 31 three-inch photomultiplier tubes (PMTs) housed in a hybrid Digital Optical Module (hDOM). We describe the system architecture, control logic, and laboratory test procedures, and report the combined PMT--base performance in terms of baseline stability, gain uniformity, and timing accuracy under conditions designed to emulate the deep-sea environment. Baseline measurements show low and stable electronic noise. Gain calibrations based on single-photoelectron spectra demonstrate that all PMTs can be tuned to a common nominal gain and remain stable over multi-day operation. Transit-time-spread measurements yield values below 1.8~ns (FWHM), consistent with manufacturer specifications. These results indicate that the CW-based HV system provides the stability and timing precision required for deep-sea multi-PMT optical modules.

hep-ex↗
arXiv2026

Assessing the validity of the Born-Oppenheimer approximation in potential models for doubly heavy hadrons

The Born-Oppenheimer approximation is widely used to investigate the properties of hydrogen-like systems and doubly heavy hadrons. However, the extent to which this approximation captures the features of such systems within potential models remains an open question. In this work, we adopt the results obtained with the Gaussian expansion method as a benchmark to assess the validity of the Born-Oppenheimer approximation within potential models for hadronic systems. We also investigate the dependence of the Born-Oppenheimer approximation results on the choice of trial wave functions. A comprehensive study of the Born-Oppenheimer approximation is carried out by performing calculations using Slater-type functions and Gaussian-type functions as trial wave functions, and by comparing the resulting predictions with those obtained from the Gaussian expansion method. We find that the calculations performed within the Born-Oppenheimer approximation are close to those obtained with the Gaussian expansion method when the heavy-quark mass is relatively small. However, as the heavy-quark mass increases, calculations employing Slater-type functions yield larger values than those from the Gaussian expansion method, whereas those using Gaussian-type functions lead to smaller ones. The use of Slater-type functions generally leads to an enhanced binding energy. The underestimation observed in Born-Oppenheimer approximation calculations with Gaussian-type functions primarily stems from the neglect of non-adiabatic corrections. This comparative study provides deeper insight into the structure of doubly heavy hadrons and helps clarify the applicability and limitations of the Born-Oppenheimer treatment within potential models.

hep-ph↗
arXiv2026

Strix: Re-thinking NPU Reliability from a System Perspective

DNNs and LLMs increasingly rely on hardware accelerators, including in safety-critical domains, while technology scaling and growing model complexity make hardware faults more frequent. Existing system-level mechanisms typically treat the NPU as a monolithic unit, using coarse-grained replication that incurs prohibitive performance and hardware overheads, leaving a gap between reliability requirements and deployable solutions. To bridge this gap, we present Strix, a full-stack NPU reliability framework on an open-source SoC, spanning micro-architecture, ISA, and programming methods. Strix re-partitions the NPU along the system inference pipeline, identifies dominant failure modes, and attaches targeted safeguards, achieving sub-micro-second fault localisation, error detection, and correction with only 1.04$\times$ slowdown and minimal hardware overhead.

cs.AR↗
arXiv2026

From Characterization to Microarchitecture: Designing an Elegant and Reliable BFP-Based NPU

Block Floating-Point (BFP) is emerging as an attractive data format for edge Neural Processing Units (NPUs), combining wide dynamic range with high hardware efficiency. However, its behavior under hardware faults and suitability for safety-critical deployments remain underexplored. Here, we present the first in-depth empirical reliability study of BFP-based NPUs. Using RTL-level fault injection on NPUs, our bit- and path-level analysis reveals pronounced heterogeneous vulnerabilities and shows conventional end-to-end check becomes ineffective under nonlinear block scaling. Guided by these insights, we design a fault-tolerant BFP-based NPU microarchitecture that aligns the BFP computational semantics with reliability constraints. The design uses a row/column-wise blocking strategy to decouple the fixed-point mantissa computations from the scalar exponent path, and introduces ultra-lightweight protection mechanisms for each. Experimental results demonstrate our design achieves near-dual modular redundancy reliability with only $3.55\%$ geometric mean performance overhead and less than $2\%$ hardware cost.

cs.AR↗
arXiv2026

Image Diffusion Preview with Consistency Solver

The slow inference process of image diffusion models significantly degrades interactive user experiences. To address this, we introduce Diffusion Preview, a novel paradigm employing rapid, low-step sampling to generate preliminary outputs for user evaluation, deferring full-step refinement until the preview is deemed satisfactory. Existing acceleration methods, including training-free solvers and post-training distillation, struggle to deliver high-quality previews or ensure consistency between previews and final outputs. We propose ConsistencySolver derived from general linear multistep methods, a lightweight, trainable high-order solver optimized via Reinforcement Learning, that enhances preview quality and consistency. Experimental results demonstrate that ConsistencySolver significantly improves generation quality and consistency in low-step scenarios, making it ideal for efficient preview-and-refine workflows. Notably, it achieves FID scores on-par with Multistep DPM-Solver using 47% fewer steps, while outperforming distillation baselines. Furthermore, user studies indicate our approach reduces overall user interaction time by nearly 50% while maintaining generation quality. Code is available at https://github.com/G-U-N/consolver.

cs.LG↗