arXiv subjects

Liang Zhang

Publications and source records attributed to Liang Zhang.

At least 19 recordsLinked to original sources

arXiv2026

EmoPose: Vision-Language Model Guided Emotion-Aware Gesture Generation for Humanoid Robots

Socially competent humanoid robots must communicate affect and intent through gesture as well as speech, yet open-ended interaction must become motion that is both expressive and executable on a specific body. This demands semantic flexibility for contextual social intent while preserving deterministic, embodiment-aware robot control. We present EmoPose, a vision-language model (VLM)-guided framework that bridges this gap through an executable semantic interface. Given language, dialogue history, and optional visual context, the VLM selects an ordered gesture plan containing a communicative class, library variant, intensity, and speech anchor. A scalable robot-owned motion library defines the available expressive vocabulary and the source of 14-DoF joint targets. Pose Studio supports automatic trajectory generation, MuJoCo preview, and automatic synchronization of new library entries with the VLM guide; deterministic robot-side modules validate plans, construct trajectories, schedule gestures, and manage queueing and interruption. This division lets the interaction repertoire grow for new social contexts without changing the control interface or delegating raw joint commands to the foundation model. On the EmoPose-Bench, structured GPT-5.5 planning reaches $98.25\pm0.52\%$ on the Easy tier and $76.50\pm0.54\%$ overall, exceeding same-model direct-label prompting. Further tests validate dialogue-context use and ordered multi-action composition. The system completes the nominal MuJoCo suite and realizes all 29 authored variants on the physical Unitree G1. A four-stop laboratory tour demonstrates expressive narration with interruption, camera-grounded dialogue, and navigation.

cs.RO↗

arXiv2026

Dynamics of Weighted Backward Shifts on Cesàro Spaces of Rooted Trees

We study the dynamics of weighted backward shifts on Ces`aro spaces associated with leafless locally finite rooted trees. We first characterize their boundedness in terms of adjacent level cardinalities and edge weights. We then characterize their $\mathcal{F}$-transitivity by a growth condition involving level cardinalities, products of weights along paths, and a level-dependent Ces`aro factor. As consequences, we obtain criteria for hypercyclicity, weak mixing, topological ergodicity, and topological mixing. We also characterize the existence of nonzero orbit limit points and chaotic weighted shifts, the latter in terms of normalized fixed points and unit flows satisfying an explicit summability condition. Examples show that $\mathcal{F}_{\underline{d}>0}$-transitivity need not imply frequent hypercyclicity, that a nonhypercyclic weighted shift may nevertheless have a nonzero orbit limit point, and that topological mixing need not imply chaos.

math.DS↗

arXiv2026

LiteViLNet: Lightweight Vision-LiDAR Fusion Network for Efficient Road Segmentation

Road segmentation is a fundamental perception task for autonomous driving and mobile robotics, where both appearance and geometric cues must be processed under edge-computing constraints. Existing multi-modal approaches often improve accuracy with large encoders or expensive global interaction, which limits their use on embedded platforms. We present \textbf{LiteViLNet}, a lightweight RGB-geometry fusion network that combines a MobileNetV3 RGB encoder with a 0.12M-parameter depth-wise-separable geometry encoder. A multi-scale feature fusion module performs modality-specific enhancement, global-query cross-modal interaction, and adaptive gating, while a depth-wise large-kernel bridge enlarges the contextual support of the deepest representation with low overhead. The resulting U-Net-style decoder uses deep supervision only during training. On the KITTI Road benchmark, the 14.04M-parameter full model obtains $97.23\pm0.15\%$ MaxF. On the held-out ORFD test set under the released OFF-Net evaluation protocol, the full model achieves $96.74\pm0.09\%$ F-score and $93.68\pm0.18\%$ IoU. On a Jetson Orin NX, model-only PyTorch FP16 inference reaches $22.18\pm0.21$ FPS; a separate TensorRT FP16 measurement reaches $68.73\pm0.06$ FPS on the Jetson. Camera-depth adaptations and perception-and-control demonstrations on three heterogeneous robot platforms further illustrate the portability of the dual-stream design.

cs.CV↗

arXiv2026

Compelling evidence of a link between the lags of the quasi-periodic oscillations and the radio jet in the black-hole X-ray binary GRS 1915+105

GRS 1915+105 is one of the most studied black-hole X-ray binaries, known for its extreme variability and rich phenomenology. Previous studies of this source with the Rossi X-ray Timing Explorer reported a transition of the phase-lags of type-C quasi-periodic oscillations (QPOs) from soft, where low-energy photons lag the high-energy ones, when the QPO frequency is higher than ${\sim} 2 ~{\rm Hz}$, to hard when the QPO frequency decreases below ${\sim} 2 ~{\rm Hz}$. The hard-lags of the QPO coincide with episodes of strong radio emission. We analyse NICER observations of GRS 1915+105 obtained between 2018 and 2020, during a period in which the source flux decreased steadily, and perform a detailed spectral-timing study of the detected type-C QPOs. We find a type-C QPO with frequencies in the range of ${\sim} 1.3$-$3.9 ~{\rm Hz}$, which displays soft lags and, contrary to the RXTE observations, shows no evidence of hard phase lags at frequencies below $2~{\rm Hz}$. Quasi-simultaneous AMI-LA radio observations show consistently low radio flux (${\lesssim} 5 ~{\rm mJy}$) during this period. These results appear to show that the hard QPO lags in GRS 1915+105 are linked to the presence of strong radio activity, suggesting that the relativistic jet is responsible for the hard phase lags, supporting a scenario in which QPO phase lags trace changes in coronal geometry and accretion-ejection coupling in GRS 1915+105.

astro-ph.HE↗

arXiv2026

Fine-Tuning Large Language Models for Codebook-Guided Coding of Students' Mathematics Metaphor Responses

Student-generated metaphors about mathematics can provide insights into students' attitudes, beliefs, identities, and experiences, but expert human assessment through thematic coding of these semantically complex metaphor responses is labor-intensive and difficult to scale. This study examines whether Low-Rank Adaptation (LoRA)-based supervised fine-tuning of Large Language Models (LLMs) can improve their performance on codebook-guided coding tasks for student mathematics metaphors. We utilized a human-coded corpus of 2,265 Grade 6-8 responses to food- and animal-based metaphor prompts and evaluated LLMs on two tasks: valence-intensity coding of students' affective orientations toward mathematics and thematic coding of their metaphorical framings of mathematics. Two open-weight LLMs, DeepSeek-R1 1.5B and Mistral 7B, were evaluated before and after fine-tuning and compared with two proprietary LLMs, GPT-4o mini and GPT-5 mini. Results show that fine-tuning substantially improved the performance and run-to-run reliability of the open-weight LLMs across both tasks relative to their base versions, making the fine-tuned LLMs competitive with and often outperforming the proprietary LLMs. These findings suggest the potential of fine-tuned open-weight LLMs for scalable and automated AI-assisted measurement of students' metaphor responses with competitive performance while maintaining local controllability and privacy-conscious deployment.

cs.HC↗

arXiv2026

AgentLeak: Cloning Stronger LLM Agent Capabilities onto Weaker Agents Beyond Skill Stealing

Large language model (LLM) agents increasingly achieve long-horizon tasks by combining foundation models with explicit skills and implicit procedural knowledge acquired through execution. The resulting task-solving capabilities have become valuable proprietary assets, raising a new security question: can a substantially weaker attacker-controlled agent acquire the capabilities of a stronger proprietary agent through limited black-box interaction? Existing skill-stealing attacks recover explicit skill artifacts, yet we show that artifact leakage does not necessarily transfer capability: a weaker agent may possess the same skills but still fail because it lacks procedural behaviors implicitly realized by the stronger agent. Our key insight is that the skill execution gap itself forms a leakage surface, where missing behaviors are exposed through observable differences between successful victim executions and failed attacker executions. Based on this, we present AgentLeak, a black-box capability-cloning attack that identifies capability-critical behaviors from these execution differences and incorporates them into attacker-side skills, while keeping the attacker's model, harness, and tools unchanged. Across 20 task scenarios comprising 600 instances, diverse agent systems, and multiple backbone models, AgentLeak improves task pass rates by over 40% compared with direct skill reuse and recovers more than 80% of the victim--attacker capability gap. Our findings reveal a confidentiality risk in LLM agents: protecting explicit artifacts alone is insufficient, as observable execution behavior can leak the procedural knowledge required to reconstruct proprietary task-solving capabilities in low-capability and attacker-controlled agents.

cs.CR↗

arXiv2026

Look Clearly Before Answering: Mitigating Hallucinations in LVLMs via Saliency-Driven Perceptual Realignment

Large vision-language models (LVLMs) have demonstrated remarkable capabilities in multimodal understanding. However, they remain prone to hallucinations, generating responses that are inconsistent with the visual evidence. Existing mitigation methods largely address language-prior bias or cross-modal imbalance, while progressive visual degradation across perception and memory remains underexplored. In this work, we propose Saliency-Driven Perceptual Realignment (SDPR), a training-free framework that mitigates the degradation of visual awareness throughout inference. Specifically, we first introduce saliency-driven attention redistribution to release attention hijacked by non-semantic sink tokens, thereby recovering critical visual evidence. Second, we identify spatial distortion in the KV cache and propose saliency-driven cache alignment to preserve query-relevant visual features during generation. Finally, we introduce prior-constrained contrastive decoding to penalize unfaithful predictions induced by dominant language priors. Our proposed SDPR is robust against hallucinations due to its holistic alignment of visual awareness across the entire generative trajectory. Extensive experiments across diverse LVLM architectures show that SDPR outperforms state-of-the-art methods on both hallucination and general-purpose benchmarks, requiring no additional training and incurring minimal runtime overhead. The code is available \href{https://github.com/PengSyuChen/SDPR}{\color{blue}{here}}.

cs.CV↗

arXiv2026

Error-Aware Reverse Auction Mechanism for Large Language Model Routing

Routing each query to a cost-effective large language model (LLM) is critical for balancing quality and cost, yet most routers rely on a centralized task center to predict model performance, creating an information-risk mismatch and a scalability bottleneck as the model pool grows. We propose a market-based routing paradigm that shifts ex-ante prediction to LLM providers via a reverse auction, where providers bid with self-predicted success probabilities and execution costs. To account for inherently noisy provider predictions and center evaluations, we introduce the \textit{\textbf{E}rror-\textbf{A}ware \textbf{R}everse \textbf{A}uction \textbf{M}echanism} (EA-RAM), which explicitly models this inherent Dual Error. We prove that EA-RAM is Bayesian incentive compatible and individually rational under the Dual Error, establish sufficient conditions for center rationality, and derive an explicit welfare-loss bound. We further identify robustness effects: opposite-signed errors can cancel, vanishing-tail link functions (e.g., logistic) stabilize clear-cut cases via saturation, and extra noise smooths belief maps, reducing the gains from marginal manipulation. Experiments on simulations and real-world benchmarks show that EA-RAM is robust to the Dual Error and achieves a better cost--performance Pareto frontier than centralized baselines, with additional gains when providers contribute local information, validating its practical effectiveness.

cs.GT↗

arXiv2026

ERSkill: Evolving for Skill-Guided Adaptive Memory Retrieval

While Large Language Model (LLM) agents increasingly rely on long-term memory for persistent interactions, the retrieval mechanisms governing this memory are rarely treated as evolvable components. This static approach limits performance on heterogeneous memory queries, which often demand diverse evidence construction strategies. To address this, we introduce \textbf{ERSkill}, a retrieval-centric framework for self-evolving, skill-guided memory access. ERSkill compiles interaction histories into a structured memory store and represents retrieval behaviors as executable skills composed of fundamental primitives. At inference time, a trained router dynamically matches each query to the optimal skill to construct tailored evidence for answer generation. To enable continuous improvement, ERSkill co-evolves the skill set and the router during training. It employs an experience trie to efficiently record explored retrieval paths, alongside a double-frontier mechanism that safely decouples the expansion of new skill capabilities from stable, router-facing deployment. Experiments across multiple agent memory benchmarks demonstrate that ERSkill substantially outperforms strong non-evolving and self-evolving baselines. Notably, it improves the overall average across F1, BLEU-1, and LLM-judge scores by 31.3\% with Qwen3-Next-80B-A3B-Instruct and by 28.1\% with GPT-5.4-nano.

cs.CL↗

arXiv2026

Cross-spectral Analysis of the Type-C Quasi-periodic Oscillation Shoulder Component in GX 339-4

We revisit Rossi X-ray Timing Explorer (RXTE) observations of GX~339$-$4 during the rising phase of its 2006/2007 outburst and apply a joint power-density-spectrum (PDS)--cross-spectrum (CS) decomposition to the type-C quasi-periodic oscillation (QPO) region. Within this framework, the QPO region is described by a narrow QPO fundamental and a neighboring high-frequency shoulder, whose amplitudes and phase lags can be measured separately. The shoulder is first detected at MJD~54142.04, mainly through the imaginary part of the CS and a narrow local structure in the phase-lag spectrum, before becoming a resolved high-frequency shoulder in the PDS. It follows the QPO frequency evolution on the high-frequency side, with $R_ν=ν_{\rm sh}/ν_{\rm QPO}\simeq1.04$--$1.18$. The QPO lag remains small, typically below $\sim0.17$~rad, whereas the shoulder carries a larger hard lag of $\sim0.5$--$0.8$~rad. Energy-resolved fits show the same separation: the QPO lag is close to zero or only weakly positive across most of the energy band, while the shoulder lag is systematically larger and generally increases with photon energy. The two components have broadly similar rms--energy shapes, although their relative strengths evolve during the observed sequence. Although the shoulder remains broad, with $Q\sim2$--$4$, its lag and rms--energy behavior resemble those of the type-B QPO detected shortly after our observations. This similarity raises the interesting possibility that the shoulder is related to an earlier, broader stage of the variability later seen as the type-B QPO.

astro-ph.HE↗

arXiv2026

Phase-resolved QPO Analysis of GX 339-4: Improved Technique and Consistent Behaviors between QPOs and Broadband Noise

The nature of low-frequency quasi-periodic oscillations (QPOs) in black hole X-ray binaries remains unclear, and their relationship with the accompanying broadband noise (BBN) is still under debate. Here, we propose an improved variational mode decomposition (VMD) technique. Compared with the original algorithm that requires iterative, case-by-case parameter tuning, the new algorithm automatically and consistently determines the relevant VMD parameters based on the QPO central frequency and width measured from the power spectral density (PSD). This enables a more robust phase determination for QPOs and can also be applied to the study of BBN. We found that, for low-frequency type-C QPOs without significant harmonics in the black hole X-ray binary GX 339-4, the spectral properties of the QPOs and BBN are statistically consistent with each other: (1) the photon index is positively correlated with count rate as a function of phase, and (2) the PSD ratio spectra across different energy bands show no statistically significant QPO-like structures near the QPO frequencies, indicating that both components share the same energy dependence. These suggest that QPOs and BBN may be driven by the same physical processes. The QPO models based on geometric modulation struggle to account for the results, while those invoking corona oscillations are favored.

astro-ph.HE↗

arXiv2026

Don't Peek at the Answer: Outcome-Masked Group Relative Policy Optimization for Label-Free RLVR

Reinforcement Learning with Verifiable Rewards (RLVR) improves LLM reasoning but typically relies on ground-truth (GT) answers, limiting scalability. Voting-based label-free RLVR replace gold supervision with answer-level consensus from model samples. However, collapse arises when the same answer-level signal is used both to estimate rewards and to drive token-level policy optimization, encouraging the model to directly reinforce answer tokens rather than improve reasoning. We propose OM-GRPO, a label-free RLVR framework that decouples reward estimation from policy optimization. OM-GRPO masks gradients on the answer span while retaining answer-level rewards through a soft consensus signal, shifting optimization pressure away from answer tokens. We further introduce Contrast-Augmented Reward, which refines reward estimation via low-cost pairwise comparisons over existing trajectories without additional rollouts. Across diverse reasoning benchmarks and three LLM backbones, OM-GRPO consistently outperforms existing label-free RLVR methods and matches supervised GT-reward training with stable optimization. This stability is particularly beneficial in the Test-Time Training setting, where OM-GRPO surpasses majority voting by 4.24 points.

cs.AI↗

arXiv2026

Knowledge-guided Disentanglement with Atomic Actions for Action Recognition

Action recognition in complex scenes often involves multiple concurrent fine-grained actions, making it challenging to model internal action structures. Most existing methods rely on holistic representations, which are insufficient for capturing subtle interactions and fine-grained semantics. While recent prompt-based approaches introduce disentanglement, they lack explicit semantic guidance, and methods based solely on visual or structured cues remain coarse-grained. In this paper, we propose Knowledge-guided Disentanglement with Atomic Actions (KDA), which leverages fine-grained semantic knowledge to enhance action representations and enable more precise disentanglement. Specifically, we use Large Language Models (LLMs) to decompose action labels into atomic actions, providing explicit spatial-temporal semantics. A Knowledge Injection Module (KIM) first integrates atomic action knowledge into video features. Based on this enhanced representation, a Knowledge Disentanglement Module (KDM) further disentangles atomic action knowledge to produce more precise semantic guidance for action disentanglement. A Knowledge Disentanglement Loss (KD Loss) is introduced to encourage clearer disentanglement of knowledge components within KDM. Extensive experiments demonstrate that KDA improves feature discriminability and achieves state-of-the-art performance on multi-label action recognition benchmarks. Moreover, KIM and KDM can be readily integrated into other methods, demonstrating strong generality.

cs.CV↗

arXiv2026

CTR-Sink: Attention Sink for Language Models in Click-Through Rate Prediction

Click-Through Rate (CTR) prediction, a core task in recommendation systems, estimates user click likelihood using historical behavioral data. Modeling user behavior sequences as text to leverage Language Models (LMs) for this task has gained traction, owing to LMs' strong semantic understanding and contextual modeling capabilities. However, a critical structural gap exists: user behavior sequences consist of discrete actions connected by semantically empty separators, differing fundamentally from the coherent natural language in LM pre-training. This mismatch causes semantic fragmentation, where LM attention scatters across irrelevant tokens instead of focusing on meaningful behavior boundaries and inter-behavior relationships, degrading prediction performance. To address this, we propose $\textit{CTR-Sink}$, a novel framework introducing behavior-level attention sinks tailored for recommendation scenarios. Inspired by attention sink theory, it constructs attention focus sinks and dynamically regulates attention aggregation via external information. Specifically, we insert sink tokens between consecutive behaviors, incorporating recommendation-specific signals such as temporal distance to serve as stable attention sinks. To enhance generality, we design a two-stage training strategy that explicitly guides LM attention toward sink tokens and a attention sink mechanism that amplifies inter-sink dependencies to better capture behavioral correlations. Experiments on one industrial dataset and two open-source datasets (MovieLens, Kuairec), alongside visualization results, validate the method's effectiveness across scenarios. The source code of this paper has been made publicly available at https://github.com/UGUESS-lzx/CTR-SINK.

cs.CL↗

arXiv2026

X-ray and Radio campaign of the Z-source GX 340+0: discovery of X-ray polarization and its implications

We present the discovery of X-ray polarization from the neutron star low-mass X-ray binary and Z-source, GX~340$+$0, using an Imaging X-ray Polarimetry Explorer (IXPE) observation in March 2024. Along with the IXPE observation, we conducted an extensive X-ray and radio monitoring campaign to ascertain the source properties during and around the IXPE observation. The source was within the horizontal branch throughout the multiwavelength campaign. We measured a significant X-ray polarization in 2--8 keV with polarization degree (PD) = $4.02 \pm 0.35$% and polarization angle (PA) = $37.6 \pm 2.5^\circ$. The energy-dependent polarization indicates that in the 2-2.5 keV energy range, the PA is much lower, $\sim9\pm8^\circ$, while other energy bands are consistent with the PA found over 2.5--8 keV. The simultaneous AstroSat-IXPE spectro-polarimetric observations provide some evidence for independent polarization from various spectral components, hinting at a disparity in the PA from the accretion disk and the Comptonized emission, while suggesting an unpolarized emission from the blackbody component. Radio observations in the 0.7--9 GHz frequency range reveal a non-detection of radio emission in 0.7-1.5 GHz and a significant detection in 5.5--9 GHz, suggesting the presence of a spectral break in 1.5-5.5 GHz. Using ATCA observation we place upper limits on the radio polarization at $<$6% on the linear polarization and $<$4% on the circular polarization at 3$σ$ level. We discuss the origin of the X-ray polarization and its implications on the geometry of the spectral components.

astro-ph.HE↗

arXiv2026

Tensor Reduction of Sunset by Generating Function

Recently, the generating function has been proposed as an alternative reduction method. This method has been tested at the one-loop level, including the tensor reduction and propagators with higher powers. In this work, we initiate the study of the method for higher loops by focusing on the sunset diagram, which is the simplest nontrivial two-loop integral. By employing PV reduction equations together with syzygy equations, we construct a complete system of differential equations. Through series expansion, we derive a complete set of recurrence relations, which can efficiently reduce any high-rank tensor structure.

hep-th↗

arXiv2026

Resonance Femtoscopy Beyond the On-Shell Approximation

The observed shift of the $Δ(1232)$ resonance peak in $π$-$p$ femtoscopic correlations challenges the conventional Breit-Wigner description of resonances in femtoscopy. We revisit the Koonin-Pratt framework by formulating femtoscopy in the momentum-space representation. By employing the T-matrix approach to disentangle on-shell and off-shell contributions, we show that the finite spatial extent of the emission source naturally induces sensitivity to off-shell scattering dynamics. Using a Friedrichs-Lee model constrained by low-energy $π$-$p$ scattering data, we numerically demonstrate that this off-shell sensitivity leads to a peak shift accompanied by a dip on the high-momentum side of the peak. The predicted high-momentum side dip is absent in the data, pointing to source properties beyond the simple Gaussian approximation.

nucl-th↗

arXiv2026

Fast Optical Variability of the TeV Blazar PKS 1725+123 Observed by SVOM-VT and Insights from Multi-wavelength Follow-up Observations

PKS 1725+123 is a flat-spectrum radio quasar (FSRQ) with a redshift of $z=0.586$. The detection of this object in the TeV band was reported by the MAGIC telescopes and H.E.S.S. in August 2025. Subsequently, we promptly initiated Target-of-Opportunity observations using the Space-based multi-band astronomical Variable Objects Monitor (SVOM) satellite. By analyzing the observational optical data from SVOM-VT and comprehensively examining the Fermi-LAT and Swift-XRT observational data, it was found that the source is in a high-flux state across the optical, X-ray, and GeV $γ$-ray bands around the time of the TeV detections. Its optical flux reaches a historically unprecedented high level and shows significant variability on timescale as short as minutes. The variability is accompanied by changes in the color index, exhibiting a bluer when brighter behavior during the high-flux state. Based on the simultaneous multi-wavelength data, we construct the broadband spectral energy distribution (SED) of the source in the high-flux state. PKS 1725+123 demonstrates a remarkably high synchrotron peak frequency, which is distinctly different from that of other FSRQs. We propose a two-zone spine-sheath jet model to reproduce this SED. The optical--X-ray emission is generated by the synchrotron process of the relativistic electrons within a compact zone. The inverse Compton (IC) scattering processes of the same electron population contribute to the low-energy end of the Fermi-LAT spectrum, while the high-energy end of the Fermi-LAT spectrum is ascribed to the IC scattering of the synchrotron photons within the compact zone by the higher-energy electrons in an extended region.

astro-ph.HE↗