arXiv subjects

Jun Luo

Publications and source records attributed to Jun Luo.

At least 19 recordsLinked to original sources

arXiv2026

Visual Attention Faithfulness in Vision-Language Models is Heterogeneous

Whether attention weights faithfully reflect model reasoning has been actively debated in NLP, yet this question remains largely unexplored for the visual modality in Vision-Language Models (VLMs). We address this gap through causal perturbation analysis on current VLMs, evaluating both the comprehensiveness and sufficiency gap of attention-ranked visual tokens. Our analysis reveals that visual attention faithfulness is heterogeneous, manifesting in three distinct processing modes: Faithful-Sufficient, where top-$k$ attention tokens are both necessary and sufficient for prediction; Faithful-Distributed, where they are necessary but broader visual context remains required; and Non-Focal, where no localized attention region is individually necessary while visual information remains an essential trigger for prediction. Furthermore, human-annotated ground-truth regions satisfy comprehensiveness in only $\sim 60$% of cases compared with model attention rankings, revealing systematic divergence between model visual reliance and human intuition. We demonstrate these patterns across both general VQA on VQAv2 and document tasks on VRDU and ChartQA, showing that visual attention faithfulness varies systematically with processing demands and model architectures rather than being uniformly faithful or unfaithful.

cs.CV↗

arXiv2026

Beyond Token-Level Guidance: Inference-Time Alignment of Specialized LLMs via Cross-Family Representation Steering

Large language models (LLMs) finetuned for specialized domains represent crucial high-impact applications. Inference-time alignment improves safety degraded from specialization finetuning without requiring substantial computational resources, complementing finetuning-based methods with an easy-to-use, plug-and-play solution. However, existing inference-time methods fail to reliably improve safety without disrupting domain capability. We identify the root cause as complementary expertise orthogonality: specialized base models and general-domain guidance models have orthogonal competencies, making the guidance signal unreliable for specialized generation. This primarily manifests as stop token interference, where the guidance model's tendency toward continuation overrides the base model's decision to stop, burying correct answers under guidance-induced continuation. To address this problem, we propose CREST, an inference-time alignment method that steers base model hidden representations using safety directions extracted from a guidance model of any family, avoiding token-level structural limitations entirely. CREST improves safety where specialization has weakened it while preserving both domain-specific capability and the safety of already well-aligned models, outperforming baselines by up to 22.2\% on safety benchmarks. Our code is available at: https://github.com/DecayingSeart/CREST.

cs.CL↗

arXiv2026

TIGA: Trajectory-Injected Generative Attack against Black-box AIGC Detectors

Recent diffusion models have achieved remarkable realism in facial image synthesis, posing growing challenges to artificial intelligence-generated content (AIGC) forensic detectors.Existing evasion methods typically perturb pre-generated images or require detector-aware training, which may introduce visible or statistical artifacts and limit applicability when the diffusion model must remain frozen and the target detector is accessible only through black-box queries. We propose Trajectory-Injected Generative Attack (TIGA), a source-image-free and training free framework that generates detector-evasive images within a single diffusion sampling trajectory. TIGA steers the latent Denoising Diffusion Implicit Model (DDIM) trajectory so that adversarial properties emerge during generation rather than being added afterward. TIGA first aggregates gradients from multiple white-box surrogate detectors to form a transferable, sign-aware prior, and then performs anisotropic directional search with symmetric finite-difference queries to estimate the black-box target response. The estimated directions are stabilized by decayed momentum and injected according to the DDIM noise schedule, with frequency-domain reshaping to suppress high frequency artifacts. Experiments on surrogate and unseen specialized forensic detectors show that TIGA achieves strong blackbox attack performance, transferability, and high robustness under common post-processing operations without source images or diffusion-model retraining, while preserving high perceptual quality.

cs.CV↗

arXiv2026

The Sample Pre-selection and Characterization Station at the SECUF: Instrumentation, Capabilities, and Representative Scientific Achievements

The Synergetic Extreme Condition User Facility (SECUF) is a comprehensive, state-of-the-art user facility designed to provide integrated extreme physical conditions-including ultrahigh pressure, ultralow temperature, strong magnetic fields, and ultrafast optical fields-for frontier research in condensed matter physics and materials science. Within SECUF, the F2 Sample Pre-selection and Characterization Station plays a pivotal supporting role. Its mission is to provide comprehensive sample synthesis, processing, pre-screening, and characterization services to prepare high-quality specimens for subsequent experiments under extreme conditions. This paper details the specifications and performance of ten core instrument systems within these units. Furthermore, we highlight several breakthrough scientific achievements enabled by the F2 Station, encompassing the discovery of novel quantum spin supersolid states, pressure-induced high-temperature superconductivity in nickelates, giant anomalous Hall angles, and molecular water in lunar soil. We also outline ongoing technical developments that expand the station's capabilities, such as integrated high-pressure cells and self-built ancillary measurement systems.

cond-mat.str-el↗

arXiv2026

Precision Recall Controllable Radiology Report Generation via Hybrid Natural Language and Clinical Reward Learning

Automated radiology report generation (RRG) has gained increasing attention because it can reduce the heavy workload of clinical report writing. However, most existing methods mainly optimize for natural language generation (NLG) metrics that focus on language fluency, while providing little control over clinically important factors such as precision and recall. As consequence, generated reports may be fluent but not well aligned with different clinical needs. To address this challenge, we propose a reinforcement learning framework for precision recall controllable RRG, where a control parameter explicitly adjusts the trade-off between clinical precision and recall during inference. This design allows the model to flexibly generate reports according to different clinical requirements. To ensure clinical correctness, we introduce a clinical reward into the training objective, which helps improve clinical efficacy (CE) beyond standard language-based optimization. In addition, we apply a group-relative training strategy that normalizes rewards within each training group, reducing reward variance and improving training stability. Extensive experiments on the MIMIC-CXR dataset show that our method consistently outperforms state-of-the-art approaches in both NLG and CE evaluation metrics, while providing reliable control over the CE precision recall trade-off.

cs.CL↗

arXiv2026

PACT: Privileged Trace Co-Training for Multi-Turn Tool-Use Agents

Multi-turn tool-use agents must reason, call tools, and adapt to observations across several interaction turns. Post-training such agents is challenging, as reinforcement learning often suffers from sparse rewards and weak credit assignment despite matching the prompt-only inference setting, while supervised fine-tuning on expert traces provides dense process supervision but can over-constrain the model to fixed trajectories. To tackle this, we propose PACT, a Privileged trAce Co-Training framework for multi-turn tool-use agents. The key idea is to use expert traces only as training-time optimization signals rather than rollout-time hints. PACT keeps rollout generation prompt-only, then uses expert traces to guide optimization through two complementary signals: a trace-conditioned RL surrogate that evaluates prompt-only rollouts under expert-trace context, and a component-aware SFT loss that supervises reasoning prefixes and tool-calls with annealed strength. To reduce over-reliance on the training-only trace context, PACT further introduces a prompt-only anchoring. We also provide a latent-trace view that connects the two trace-based objectives and explains how expert traces can guide optimization without being used during rollout generation. Experiments on FTRL, BFCL, and ToolHop show that PACT consistently improves over strong SFT- and RL-based baselines, highlighting the value of privileged trace co-training for multi-turn tool-use learning.

cs.CL↗

arXiv2026

Conflict-Aware Federated Fine-Tuning of Large Language Models with Mixture-of-Experts

The continuous scaling of large language models (LLMs) incurs prohibitive computational costs, making Mixture-of-Experts (MoE) a scalable alternative for efficient fine-tuning via sparse activation. While federated learning (FL) emerges as the paradigm for privacy-preserving collaborative optimization, integrating MoE into FL under data heterogeneity may trigger conflicting expert optimizations. Client-specific data distributions force same-indexed experts to optimize under inconsistent or even conflicting feature-label correlations. This mismatch induces destructive interference during aggregation, thus destabilizing the optimization trajectory and degrading model performance. To address this issue, we propose FC-MoE, a federated conflict-aware framework for MoE fine-tuning. It employs an importance aware weighting scheme to prioritize reliable local updates and utilizes gradient consensus projection to suppress conflicting updates, ensuring a stable global optimization path. Moreover, a local knowledge retention mechanism further preserves specialized client expertise by re-anchoring domain-specific residuals. Extensive experiments demonstrate that FC-MoE accelerates convergence and enhances both global and local model performance in non-IID federated environments.

cs.LG↗

arXiv2026

Ultralow shot noise limited giant passive resonant gyroscope for Earth rotation measurement

Optical gyroscopes directly measure the Earth's rotation and are promising instruments for real-time geophysical observations and Earth orientation parameter (EOP) determination requiring both high precision and high temporal resolution. Large-scale ring laser gyroscopes (RLGs) currently reach rotational resolutions around $10^{-11}\,\mathrm{(rad/s)/\sqrt{Hz}}$, but their quantum noise limits make it challenging to meet the requirements of future high-temporal-resolution EOP measurements. Passive resonant gyroscopes (PRGs), on the other hand, offer a potentially lower photon shot noise limit and more flexible power scaling, even if their demonstrated rotational resolutions are still about two orders of magnitude below those of leading RLGs. Here we demonstrate a $64\,\mathrm{m^{2}}$ giant passive resonant gyroscope HUST-2, and develop with an extremely low shot noise level. We experimentally obtain a shot noise limited of $5.7(1)\times10^{-13}\,\mathrm{(rad/s)/\sqrt{Hz}}$ at $1\,\mathrm{mW}$ incident optical power, following the characteristic $1/\sqrt{P}$ scaling. Through systematic suppression of dominant technical noise sources, HUST-2 further achieves a measured rotational resolution of $3\times10^{-11}\,\mathrm{(rad/s)/\sqrt{Hz}}$, bringing PRGs into the performance regime of leading large-scale RLGs for the first time. The gap between the present demonstrated rotational resolution and the shot noise limit indicates nearly two orders of magnitude further improvement potential. Reaching this limit would enable high-precision length-of-day (LOD) measurements with $10$-$100\,\mathrm{s}$ temporal resolution and lays the foundation for future large-scale gyroscope networks dedicated to real-time EOP determination.

physics.optics↗

arXiv2026

FGRPO: Federated GRPO with Adaptive Aggregation on Non-IID Data

Recent advances in language models have established reinforcement learning as the primary paradigm for eliciting self-correction and long-chain reasoning. While group relative policy optimization (GRPO) offers superior scalability by eliminating the critic network, deploying it on a central infrastructure entails collecting a large volume of data from distributed owners, which poses significant privacy risks. To address these concerns, we introduce federated GRPO (FGRPO), a framework designed to decentralize the fine-tuning of reasoning models across heterogeneous data owners. To effectively mitigate the instability caused by divergent reward scales across heterogeneous tasks, FGRPO incorporates an adaptive aggregation mechanism based on relative performance gain. By characterizing each client's improvement relative to its personalized historical baseline, the framework dynamically prioritizes effective learning trajectories regardless of local task difficulty. FGRPO ensures robust convergence on non-IID data while preserving data privacy.

cs.LG↗

arXiv2026

DECA: Decentralizing Block-Wise Adam for Efficient LLM Full-Parameter Fine-Tuning on Non-IID Data

Fine-tuning large language models (LLMs) in privacy-sensitive and resource-constrained environments remains challenging. Since training data are often distributed across multiple clients, decentralized fine-tuning offers a natural paradigm for collaborative adaptation without a central server. However, enabling full-parameter fine-tuning (FPFT) in this decentralized setting is difficult: FPFT provides strong adaptation capacity but incurs prohibitive resource consumption for billion-scale models. Existing decentralized LLM fine-tuning methods therefore mainly rely on parameter-efficient updates, which improve efficiency but may restrict downstream performance. Moreover, client data are typically non-IID, making decentralized optimization more vulnerable to client drift and unstable convergence. To address these challenges, we propose DECA, a resource-efficient decentralized FPFT framework for LLMs on non-IID data. DECA partitions model parameters into disjoint blocks and performs sequential block-wise Adam optimization, reducing resource consumption while preserving decentralized full-parameter adaptation. To stabilize training, DECA further introduces first- and second-order block-wise moment estimates with fresh local gradient statistics and consensus-derived discrepancy signals. We provide rigorous theoretical analysis and extensive experiments, showing that DECA achieves fast convergence, strong downstream performance, and significant resource efficiency.

cs.LG↗

arXiv2026

What Makes LVLMs Hallucinate Less? Unveiling the Architectural Factors Behind Hallucination Robustness

Hallucination remains one of the key challenges undermining the reliability of Large Vision-Language Models (LVLMs). But what makes an LVLM hallucinate less? Many existing efforts focus on improving internal components of the model. We argue that hallucination fundamentally stems from how the model architecture is designed. To investigate this, we factor the architecture design into three dimensions: Linguistic Foundation (LF), Visual Representation (VR), and Semantic Alignment (SA), and categorize hallucinations into Co-occurrence, Similarity, and previously overlooked Uncertainty types. Building on this formulation, we propose CoSimUE, a benchmark that creates fine-grained hallucination scenarios through controlled textual perturbations and random perturbations, enabling mapping between design choices and hallucination behaviors. Experiments across 7 design aspects show that: 1) the widely emphasized scaling of model parameters has only limited impact on reducing all three types of hallucinations; 2) larger and better-trained language foundations can reduce co-occurrence hallucinations; 3) stronger visual encoders and higher resolutions mitigate similarity errors; 4) effective alignment strategies alleviate uncertainty hallucinations. 5) Furthermore, cross-dimensional analysis reveals that jointly enhancing visual fidelity and alignment quality yields the most comprehensive improvements. This study provides the first systematic exploration linking architecture-level design to hallucination robustness, offering practical guidance for developing reliable and efficient LVLMs.

cs.CV↗

arXiv2026

Design and Characterization of Racetrack 3D-Trench Silicon Sensor Based on 8-Inch Process with Excellent Time Resolution

In the extreme environments of high-luminosity colliders, traditional planar silicon sensors suffer severe radiation-induced performance degradation and fail to satisfy the stringent demands of high-precision tracking and high-speed timing in particle physics. 3D silicon sensors enhance radiation hardness by shortening charge collection distance, yet conventional designs with columnar or square-cell trench electrodes exhibit non-uniform electric fields, including saddle points and low-field regions, which degrade charge collection efficiency and timing resolution. This work presents a novel racetrack 3D-trench silicon sensor with continuous racetrack electrodes surrounding a long central collection electrode, aiming to eliminate electric field inhomogeneities. For the first time, a 23 $\mu $m shallow-etched device was fabricated on an 8-inch platform, which provides a promising basis for its subsequent mass production and engineering applications. The device performance was systematically evaluated through theoretical analysis, 3D TCAD simulations, and characterization using semiconductor parameter analyzers and transient current technique (TCT) measurements. The sensor achieves leakage current below 0.2 nA, breakdown voltage above 110 V, full depletion voltage as low as a few volts, capacitance as low as 650 fF, collected charge of 4 fC, time response of about 640 ps, and time resolution of 50 ps. This large-scale manufacturable, shallow-etched racetrack 3D-trench silicon sensor provides a competitive device solution for portable radiation detection and next-generation 4D tracking under high-radiation and high-event-rate conditions.

physics.ins-det↗

arXiv2026

Si/SiGe multi-channel superlattice structure epitaxial growth with segmented temperature control for Next-Generation Logic Devices

Stacking multiple SiSiGe channels in advanced logic devices faces severe thermal budget accumulation, which degrades interfaces via Ge-Si interdiffusion and strain relaxation.This strategy lowers the Ge diffusion coefficient to 5.6-7% of its value at 650C (Arrhenius estimate), suppressing interdiffusion and preserving pseudomorphic strain. The 4 + 4 channel stack exhibits clear XRD satellite peaks, fully coherent strain state (reciprocal space mapping), sharp interfaces (1.5-2.6 nm transition width) and low RMS roughness (0.08 nm). Quantitative analysis from bottom to top reveals that prolonged high-temperature exposure broadens bottom interfaces and dilutes Ge concentration (from 20% to 18.5%), while the top stack maintains design targets. This work provides a process-physics understanding of thermal budget effects in multi-channel superlattices and establishes a high-quality material foundation for advanced logic devices beyond 2 nm node.

cond-mat.mtrl-sci↗

arXiv2026

FluxShard: Motion-Aware Feature Cache Reuse for Collaborative Video Analytics in Mobile Edge Computing

Caching and reusing intermediate features across consecutive frames is a common technique to reduce redundant computation and transmission for edge-cloud video analytics in mobile edge computation. Existing methods manage the cache in a fixed or globally shifted coordinate system, treating it as an indivisible whole. Under the non-uniform motion patterns of mobile scenes, this whole-scene granularity invalidates large portions of the cache even when most content has merely shifted spatially, wasting computation and bandwidth. The root cause is a granularity mismatch: the cache is managed per scene, yet motion varies per region. In this paper, we present FluxShard, a motion-aware edge-cloud video analytics system that uses codec-level block motion vectors (MVs) to manage feature cache reuse and recomputation at the granularity of individual motion regions. By re-indexing cached features along per-block MVs, FluxShard separates spatial displacement from content changes, recovering reusable content that whole-scene methods would otherwise discard. To ensure correct reuse under heterogeneous motion, the Receptive Field Alignment Principle (RFAP) identifies, from the input-level MV field alone, the positions that must be recomputed due to inconsistent spatial composition within receptive fields. To maintain cache coherence across frames, MV-guided cache remapping warps the entire feature cache to the current coordinate system each frame, sustaining a high reuse ratio over time. A profiling-driven dispatcher routes the remaining sparse workload between edge and cloud for lower latency. Evaluation across multiple vision tasks, dynamic video benchmarks, and network conditions shows that FluxShard reduces latency by 32.6-83.8% and energy by 14.9-64.0% over all baselines under the prescribed accuracy budget.

cs.NI↗

arXiv2026

To Intervene or Not: Guiding Inference-time Alignment with Probabilistic Model Blending

The wide deployment of LLMs has made model alignment necessary to make newly trained models safely and effectively respond to user instructions. Among different methods, inference-time alignment is often cheaper as it intervenes (i.e., offers guidances) only during output generation. Existing proposals apply guidances extracted from certain aligned models without properly assessing their reliability. Nonetheless, our systematic evaluation reveals that guidance effectiveness varies drastically across models; since ineffective guidances lead to further confusion and thus further interventions, the resulting excessive interventions typically indicate poor performance. To make interventions more effective and thus more efficient, we introduce BlendIn, an inference-time alignment framework that shifts from binary decisions to creating hybrid distributions integrating both models' knowledge. BlendIn stabilizes inference-time alignment by performing quality-aware alignment and proportionally weighting each model's contribution based on reliability. Compared with existing works, it preserves beneficial guidance while downweighting unreliable suggestions. BlendIn provides both diagnostic signals and mitigation strategies for misaligned guidance, achieving consistent and up to 50% performance improvement on challenging model pairs. Our code is available at: https://github.com/DecayingSeart/BlendIn.

cs.LG↗

arXiv2026

Physically-Induced Atmospheric Adversarial Perturbations: Enhancing Transferability and Robustness in Remote Sensing Image Classification

Adversarial attacks pose a severe threat to the reliability of deep learning models in remote sensing (RS) image classification. Most existing methods rely on direct pixel-wise perturbations, failing to exploit the inherent atmospheric characteristics of RS imagery or survive real-world image degradations. In this paper, we propose FogFool, a physically plausible adversarial framework that generates fog-based perturbations by iteratively optimizing atmospheric patterns based on Perlin noise. By modeling fog formations with natural, irregular structures, FogFool generates adversarial examples that are not only visually consistent with authentic RS scenes but also deceptive. By leveraging the spatial coherence and mid-to-low-frequency nature of atmospheric phenomena, FogFool embeds adversarial information into structural features shared across diverse architectures. Extensive experiments on two benchmark RS datasets demonstrate that FogFool achieves superior performance: not only does it exceed in white-box settings, but also exhibits exceptional black-box transferability (reaching 83.74% TASR) and robustness against common preprocessing-based defenses such as JPEG compression and filtering. Detailed analyses, including confusion matrices and Class Activation Map (CAM) visualizations, reveal that our atmospheric-driven perturbations induce a universal shift in model attention. These results indicate that FogFool represents a practical, stealthy, and highly persistent threat to RS classification systems, providing a robust benchmark for evaluating model reliability in complex environments.

cs.CV↗

arXiv2026

Dual-Envelope Constrained Nonlinear MPC for Distributed Drive Electric Vehicles Drifting Under Bounded Steering and Direct Yaw-Moment Control

Distributed drive electric vehicles offer superior yaw moment control for autonomous drifting in extreme maneuvers. Conventional drift analysis constructs stability boundaries from open loop equilibria points and assumes a fixed envelope structure. However, coupling among control inputs reshapes the phase plane and shifts saddle point location, which can invalidate open loop envelopes when used for closed loop drifting. To address this issue, a saddle point coordinate model is established in this paper by combining a nonlinear tire model with the handling diagram and explicitly accounting for road adhesion coefficient, longitudinal velocity, front wheel steering angle, and additional yaw moment. Based on saddle point properties, an extended dual envelope framework is constructed in the phase plane of slip angle and yaw rate. Using the convergence tendency of state points toward saddle points under bounded control inputs, the outer envelope defines a recoverable set under constraints on front wheel steering angle and additional yaw moment. The inner envelope characterizes the non-drifting stability region associated with unsaturated tire forces. Finally, a nonlinear model predictive control (NMPC) controller is developed using the extended dual envelope constraint. Hardware-in-the-loop experiments show that, compared with NMPC without envelope constraints, the proposed method enables smoother convergence toward the drift saddle point, reduces the steady-state tracking errors of vehicle speed, sideslip angle, and yaw rate by 33.07%, 71.18%, and 31.27%, respectively, and decreases the peak tracking error by 63.66% under road-friction mismatch.

eess.SY↗

arXiv2026

Cascade of Spin Liquids in a Bilayer Triangular-lattice Antiferromagnet Rb_2Co_2(SeO_3)_3

In frustrated Ising magnets, classical spin liquids (CSLs) with macroscopic ground-state degeneracy can survive against conventional magnetic order, as exemplified by systems on triangular, kagome and pyrochlore lattices at zero field. Here we report the discovery of a high-field route toward spin liquids in a bilayer triangular lattice antiferromagnet, Rb$_2$Co$_2$(SeO$_3$)$_3$. We demonstrate that a cascade of CSLs -- characterized by doubly degenerate one-up-one-down local spin configurations and a residual entropy of 1/2(1-M/M_s)Rln2 per mole -- emerges through field-controlled dilution of Ising dimers. Owing to the interplay of intra- and inter-layer interactions, these CSLs are further stabilized by lattice symmetry breaking at fractional magnetization plateaus. Such field-induced spin liquids can be understood as a consequence of generalized ice rules, analogous to those governing in pyrochlore antiferromagnets. In particular, the 5/6-plateau state is a candidate quantum spin liquid. Our results thereby establish a new pathway for exploring diverse spin liquid states across both classical and quantum regimes.

cond-mat.str-el↗