arXiv subjects

Xiao Liang

Publications and source records attributed to Xiao Liang.

At least 91 records · Page 5Linked to original sources

arXiv2026

Probing Internal Dynamics of Spatiotemporal Optical Vortex Strings: Spatiotemporal Attraction and Filament Stretching

Vortex dynamics are intriguing and challenging across multiple physics fields. In optics, customized spatiotemporally structured optical fields, especially spatiotemporal optical vortices (STOV), offer the potential to tailor light via coupled space-time degrees of freedom. However, the interaction mechanisms between multiple transverse orbital angular momentum singularities within a single wave packet remain elusive. This study explores the intrinsic dynamics of a STOV with three phase singularities, observing a pronounced vortex singularity oscillation phenomena by tuning the temporal dispersion. We show that these phenomena originate from the counterintuitive spatiotemporal attractive effect between vortices, which is closely related to the singularity distance. Furthermore, the stretching into filaments and annihilation behaviors is observed by introducing antivortex in the center of the wavepacket. Experimentally, we propose a Full Interferometric Retrieval of Spatiotemporal Tomography (FIRST) method that enables the complete, single-shot capture of wave packets, with excellent agreement between theoretical predictions and experimental results. To the best of our knowledge, the dynamics of transverse spatiotemporal singularities within a single wave packet are reported here for the first time. These findings confirm the existence of interesting interactions between STOV singularities, deepen our understanding of photonics and open a new direction for investigating the complex dynamics of vortex singularities in the spatiotemporal domain.

physics.optics↗

arXiv2026

SAIL-RL: Guiding MLLMs in When and How to Think via Dual-Reward RL Tuning

We introduce SAIL-RL, a reinforcement learning (RL) post-training framework that enhances the reasoning capabilities of multimodal large language models (MLLMs) by teaching them when and how to think. Existing approaches are limited by outcome-only supervision, which rewards correct answers without ensuring sound reasoning, and by uniform thinking strategies, which often lead to overthinking on simple tasks and underthinking on complex ones. SAIL-RL addresses these challenges with a dual reward system: the Thinking Reward, which evaluates reasoning quality through factual grounding, logical coherence, and answer consistency, and the Judging Reward, which adaptively determines whether deep reasoning or direct answering is appropriate. Experiments on the state-of-the-art SAIL-VL2 show that SAIL-RL improves reasoning and multimodal understanding benchmarks at both 4B and 8B scales, achieving competitive performance against commercial closed-source models such as GPT-4o, and substantially reduces hallucinations, establishing it as a principled framework for building more reliable and adaptive MLLMs. The code will be available at https://github.com/BytedanceDouyinContent/SAIL-RL.

cs.CV↗

arXiv2026

GPD: Guided Progressive Distillation for Fast and High-Quality Video Generation

Diffusion models have achieved remarkable success in video generation; however, the high computational cost of the denoising process remains a major bottleneck. Existing approaches have shown promise in reducing the number of diffusion steps, but they often suffer from significant quality degradation when applied to video generation. We propose Guided Progressive Distillation (GPD), a framework that accelerates the diffusion process for fast and high-quality video generation. GPD introduces a novel training strategy in which a teacher model progressively guides a student model to operate with larger step sizes. The framework consists of two key components: (1) an online-generated training target that reduces optimization difficulty while improving computational efficiency, and (2) frequency-domain constraints in the latent space that promote the preservation of fine-grained details and temporal dynamics. Applied to the Wan2.1 model, GPD reduces the number of sampling steps from 48 to 6 while maintaining competitive visual quality on VBench. Compared with existing distillation methods, GPD demonstrates clear advantages in both pipeline simplicity and quality preservation.

cs.CV↗

arXiv2026

Training LLMs for Divide-and-Conquer Reasoning Elevates Test-Time Scalability

Large language models (LLMs) have demonstrated strong reasoning capabilities through step-by-step chain-of-thought (CoT) reasoning. Nevertheless, at the limits of model capability, CoT often proves insufficient, and its strictly sequential nature constrains test-time scalability. A potential alternative is divide-and-conquer (DAC) reasoning, which decomposes a complex problem into subproblems to facilitate more effective exploration of the solution. Although promising, our analysis reveals a fundamental misalignment between general-purpose post-training and DAC-style inference, which limits the model's capacity to fully leverage this potential. To bridge this gap and fully unlock LLMs' reasoning capabilities on the most challenging tasks, we propose an end-to-end reinforcement learning (RL) framework to enhance their DAC-style reasoning capacity. At each step, the policy decomposes a problem into a group of subproblems, solves them sequentially, and addresses the original one conditioned on the subproblem solutions, with both decomposition and solution integrated into RL training. Under comparable training, our DAC-style framework endows the model with a higher performance ceiling and stronger test-time scalability, surpassing CoT by 8.6% in Pass@1 and 6.3% in Pass@32 on competition-level benchmarks.

cs.CL↗

arXiv2026

A General One-Shot Multimodal Active Perception Framework for Robotic Manipulation: Learning to Predict Optimal Viewpoint

Active perception in vision-based robotic manipulation aims to move the camera toward more informative observation viewpoints, thereby providing high-quality perceptual inputs for downstream tasks. Most existing active perception methods rely on iterative optimization, leading to high time and motion costs, and are tightly coupled with task-specific objectives, which limits their transferability. In this paper, we propose a general one-shot multimodal active perception framework for robotic manipulation. The framework enables direct inference of optimal viewpoints and comprises a data collection pipeline and an optimal viewpoint prediction network. Specifically, the framework decouples viewpoint quality evaluation from the overall architecture, supporting heterogeneous task requirements. Optimal viewpoints are defined through systematic sampling and evaluation of candidate viewpoints, after which large-scale training datasets are constructed via domain randomization. Moreover, a multimodal optimal viewpoint prediction network is developed, leveraging cross-attention to align and fuse multimodal features and directly predict camera pose adjustments. The proposed framework is instantiated in robotic grasping under viewpoint-constrained environments. Experimental results demonstrate that active perception guided by the framework significantly improves grasp success rates. Notably, real-world evaluations achieve nearly double the grasp success rate and enable seamless sim-to-real transfer without additional fine-tuning, demonstrating the effectiveness of the proposed framework.

cs.RO↗

arXiv2025

Domain Adaptation in Structural Health Monitoring of Civil Infrastructure: A Systematic Review

This study provides a comprehensive review of domain adaptation (DA) techniques in vibration-based structural health monitoring (SHM). As data-driven models increasingly support the assessment of civil structures, the persistent challenge of transferring knowledge across varying geometries, materials, and environmental conditions remains a major obstacle. DA offers a systematic approach to mitigate these discrepancies by aligning feature distributions between simulated, laboratory, and field domains while preserving the sensitivity of damage-related information. Drawing on more than sixty representative studies, this paper analyzes the evolution of DA methods for SHM, including statistical alignment, adversarial and subdomain learning, physics-informed adaptation, and generative modeling for simulation-to-real transfer. The review summarizes their contributions and limitations across bridge and building applications, revealing that while DA has improved generalization significantly, key challenges persist: managing domain discrepancy, addressing data scarcity, enhancing model interpretability, and enabling adaptability to multiple sources and time-varying conditions. Future research directions emphasize integrating physical constraints into learning objectives, developing physics-consistent generative frameworks to enhance data realism, establishing interpretable and certifiable DA systems for engineering practice, and advancing multi-source and lifelong adaptation for scalable monitoring. Overall, this review consolidates the methodological foundation of DA for SHM, identifies existing barriers to generalization and trust, and outlines the technological trajectory toward transparent, physics-aware, and adaptive monitoring systems that support the long-term resilience of civil infrastructure.

eess.SP↗

arXiv2025

AdaLRS: Loss-Guided Adaptive Learning Rate Search for Efficient Foundation Model Pretraining

Learning rate is widely regarded as crucial for effective foundation model pretraining. Recent research explores and demonstrates the transferability of learning rate configurations across varying model and dataset sizes, etc. Nevertheless, these approaches are constrained to specific training scenarios and typically necessitate extensive hyperparameter tuning on proxy models. In this work, we propose \textbf{AdaLRS}, a plug-in-and-play adaptive learning rate search algorithm that conducts online optimal learning rate search via optimizing loss descent velocities. We provide theoretical and experimental analyzes to show that foundation model pretraining loss and its descent velocity are both convex and share the same optimal learning rate. Relying solely on training loss dynamics, AdaLRS involves few extra computations to guide the search process, and its convergence is guaranteed via theoretical analysis. Experiments on both LLM and VLM pretraining show that AdaLRS adjusts suboptimal learning rates to the neighborhood of optimum with marked efficiency and effectiveness, with model performance improved accordingly. We also show the robust generalizability of AdaLRS across varying training scenarios, such as different model sizes, training paradigms, base learning rate scheduler choices, and hyperparameter settings.

cs.LG↗

arXiv2025

Anatomical Region-Guided Contrastive Decoding: A Plug-and-Play Strategy for Mitigating Hallucinations in Medical VLMs

Medical Vision-Language Models (MedVLMs) show immense promise in clinical applicability. However, their reliability is hindered by hallucinations, where models often fail to derive answers from visual evidence, instead relying on learned textual priors. Existing mitigation strategies for MedVLMs have distinct limitations: training-based methods rely on costly expert annotations, limiting scalability, while training-free interventions like contrastive decoding, though data-efficient, apply a global, untargeted correction whose effects in complex real-world clinical settings can be unreliable. To address these challenges, we introduce Anatomical Region-Guided Contrastive Decoding (ARCD), a plug-and-play strategy that mitigates hallucinations by providing targeted, region-specific guidance. Our module leverages an anatomical mask to direct a three-tiered contrastive decoding process. By dynamically re-weighting at the token, attention, and logits levels, it verifiably steers the model's focus onto specified regions, reinforcing anatomical understanding and suppressing factually incorrect outputs. Extensive experiments across diverse datasets, including chest X-ray, CT, brain MRI, and ocular ultrasound, demonstrate our method's effectiveness in improving regional understanding, reducing hallucinations, and enhancing overall diagnostic accuracy.

cs.CV↗

arXiv2025

CheXPO-v2: Preference Optimization for Chest X-ray VLMs with Knowledge Graph Consistency

Medical Vision-Language Models (VLMs) are prone to hallucinations, compromising clinical reliability. While reinforcement learning methods like Group Relative Policy Optimization (GRPO) offer a low-cost alignment solution, their reliance on sparse, outcome-based rewards inadvertently encourages models to "overthink" -- generating verbose, convoluted, and unverifiable Chain-of-Thought reasoning to justify answers. This focus on outcomes obscures factual errors and poses significant safety risks. To address this, we propose CheXPO-v2, a novel alignment framework that shifts from outcome to process supervision. Our core innovation is a Knowledge Graph Consistency Reward mechanism driven by Entity-Relation Matching. By explicitly parsing reasoning steps into structured "Disease, Relation, Anatomy" triplets, we provide fine-grained supervision that penalizes incoherent logic and hallucinations at the atomic level. Integrating this with a hard-example mining strategy, our approach significantly outperforms GRPO and state-of-the-art models on benchmarks like MIMIC-CXR-VQA. Crucially, CheXPO-v2 achieves new state-of-the-art accuracy using only 5k samples, demonstrating exceptional data efficiency while producing clinically sound and verifiable reasoning. The project source code is publicly available at: https://github.com/ecoxial2007/CheX-Phi4MM.

cs.CV↗

arXiv2025

Bridging Natural Language and Formal Specification--Automated Translation of Software Requirements to LTL via Hierarchical Semantics Decomposition Using LLMs

Automating the translation of natural language (NL) software requirements into formal specifications remains a critical challenge in scaling formal verification practices to industrial settings, particularly in safety-critical domains. Existing approaches, both rule-based and learning-based, face significant limitations. While large language models (LLMs) like GPT-4o demonstrate proficiency in semantic extraction, they still encounter difficulties in addressing the complexity, ambiguity, and logical depth of real-world industrial requirements. In this paper, we propose Req2LTL, a modular framework that bridges NL and Linear Temporal Logic (LTL) through a hierarchical intermediate representation called OnionL. Req2LTL leverages LLMs for semantic decomposition and combines them with deterministic rule-based synthesis to ensure both syntactic validity and semantic fidelity. Our comprehensive evaluation demonstrates that Req2LTL achieves 88.4% semantic accuracy and 100% syntactic correctness on real-world aerospace requirements, significantly outperforming existing methods.

cs.SE↗

arXiv2025

PrediFlow: A Flow-Based Prediction-Refinement Framework for Real-Time Human Motion Prediction in Human-Robot Collaboration

Stochastic human motion prediction is critical for safe and effective human-robot collaboration (HRC) in industrial remanufacturing, as it captures human motion uncertainties and multi-modal behaviors that deterministic methods cannot handle. While earlier works emphasize highly diverse predictions, they often generate unrealistic human motions. More recent methods focus on accuracy and real-time performance, yet there remains potential to improve prediction quality further without exceeding time budgets. Additionally, current research on stochastic human motion prediction in HRC typically considers human motion in isolation, neglecting the influence of robot motion on human behavior. To address these research gaps and enable real-time, realistic, and interaction-aware human motion prediction, we propose a novel prediction-refinement framework that integrates both human and robot observed motion to refine the initial predictions produced by a pretrained state-of-the-art predictor. The refinement module employs a Flow Matching structure to account for uncertainty. Experimental studies on the HRC desktop disassembly dataset demonstrate that our method significantly improves prediction accuracy while preserving the uncertainties and multi-modalities of human motion. Moreover, the total inference time of the proposed framework remains within the time budget, highlighting the effectiveness and practicality of our approach.

cs.RO↗

arXiv2025

Beyond Pass@1: Self-Play with Variational Problem Synthesis Sustains RLVR

Reinforcement Learning with Verifiable Rewards (RLVR) has recently emerged as a key paradigm for post-training Large Language Models (LLMs), particularly for complex reasoning tasks. However, vanilla RLVR training has been shown to improve Pass@1 performance at the expense of policy entropy, leading to reduced generation diversity and limiting the Pass@k performance, which typically represents the upper bound of LLM reasoning capability. In this paper, we systematically analyze the policy's generation diversity from the perspective of training problems and find that augmenting and updating training problems helps mitigate entropy collapse during training. Based on these observations, we propose an online Self-play with Variational problem Synthesis (SvS) strategy for RLVR training, which uses the policy's correct solutions to synthesize variational problems while ensuring their reference answers remain identical to the originals. This self-improving strategy effectively maintains policy entropy during training and substantially improves Pass@k compared with standard RLVR, sustaining prolonged improvements and achieving absolute gains of 18.3% and 22.8% in Pass@32 performance on the competition-level AIME24 and AIME25 benchmarks, as well as on code generation tasks. Experiments on 12 reasoning benchmarks across varying model sizes from 3B to 32B consistently demonstrate the generalizability and robustness of SvS.

cs.CL↗

arXiv2025

Further study on the lepton mass spectra and flavor mixing with $S_{3L} \times S_{3R}$ flavor symmetry

Neutrino oscillation experiments have confirmed that neutrinos are massive particles and lepton flavors are mixed. To explain the observed lepton mass spectra and flavor mixing patterns, flavor symmetry plays a crucial and unique role. In this paper, we propose a useful symmetry-breaking scheme by applying $S_{3L} \times S_{3R} \rightarrow S_{2L} \times S_{2R} \rightarrow \emptyset$ within both charged-lepton and neutrino sectors at the mass-matrix level. For the three distinct residual subgroups $S_{2L}^{(23)} \times S_{2R}^{(23)}$, $S_{2L}^{(13)} \times S_{2R}^{(13)}$ and $S_{2L}^{(12)} \times S_{2R}^{(12)}$ under consideration, we systematically analyze the various parameterizations of the lepton mass matrices. It is shown that all the three scenarios are in good agreement with current neutrino oscillation data. Notably, within the latest best-fit values of neutrino oscillation parameters, the predicted Dirac CP-violating phase $δ$ is calculated to be $294.6^\circ$, $302.3^\circ$ and $287.0^\circ$, respectively. To further assess the viability of the model, a comprehensive numerical analysis is performed by utilizing neutrino oscillation parameters at the $3σ$ level. It is found that the allowed range of $δ$ is $281.2^\circ \rightarrow 338.7^\circ$, $287.0^\circ \rightarrow 342.2^\circ$ and $282.7^\circ \rightarrow 297.0^\circ$, all fall within its $3σ$ range. These results indicate that the proposed symmetry-breaking scheme $S_{3L} \times S_{3R} \rightarrow S_{2L} \times S_{2R} \rightarrow \emptyset$ can naturally explain the realistic lepton mass hierarchy and mixing pattern, thereby providing valuable theoretical perspectives for future research.

hep-ph↗

arXiv2025

Vision-Language-Action Models for Selective Robotic Disassembly: A Case Study on Critical Component Extraction from Desktops

Automating disassembly of critical components from end-of-life (EoL) desktops, such as high-value items like RAM modules and CPUs, as well as sensitive parts like hard disk drives, remains challenging due to the inherent variability and uncertainty of these products. Moreover, their disassembly requires sequential, precise, and dexterous operations, further increasing the complexity of automation. Current robotic disassembly processes are typically divided into several stages: perception, sequence planning, task planning, motion planning, and manipulation. Each stage requires explicit modeling, which limits generalization to unfamiliar scenarios. Recent development of vision-language-action (VLA) models has presented an end-to-end approach for general robotic manipulation tasks. Although VLAs have demonstrated promising performance on simple tasks, the feasibility of applying such models to complex disassembly remains largely unexplored. In this paper, we collected a customized dataset for robotic RAM and CPU disassembly and used it to fine-tune two well-established VLA approaches, OpenVLA and OpenVLA-OFT, as a case study. We divided the whole disassembly task into several small steps, and our preliminary experimental results indicate that the fine-tuned VLA models can faithfully complete multiple early steps but struggle with certain critical subtasks, leading to task failure. However, we observed that a simple hybrid strategy that combines VLA with a rule-based controller can successfully perform the entire disassembly operation. These findings highlight the current limitations of VLA models in handling the dexterity and precision required for robotic EoL product disassembly. By offering a detailed analysis of the observed results, this study provides insights that may inform future research to address current challenges and advance end-to-end robotic automated disassembly.

cs.RO↗

arXiv2025

Gold-Medal-Level Olympiad Geometry Solving with Efficient Heuristic Auxiliary Constructions

Automated theorem proving in Euclidean geometry, particularly for International Mathematical Olympiad (IMO) level problems, remains a major challenge and an important research focus in Artificial Intelligence. In this paper, we present a highly efficient method for geometry theorem proving that runs entirely on CPUs without relying on neural network-based inference. Our initial study shows that a simple random strategy for adding auxiliary points can achieve silver-medal level human performance on IMO. Building on this, we propose HAGeo, a Heuristic-based method for adding Auxiliary constructions in Geometric deduction that solves 28 of 30 problems on the IMO-30 benchmark, achieving gold-medal level performance and surpassing AlphaGeometry, a competitive neural network-based approach, by a notable margin. To evaluate our method and existing approaches more comprehensively, we further construct HAGeo-409, a benchmark consisting of 409 geometry problems with human-assessed difficulty levels. Compared with the widely used IMO-30, our benchmark poses greater challenges and provides a more precise evaluation, setting a higher bar for geometry theorem proving.

cs.AI↗

arXiv2025

ReMatch: Boosting Representation through Matching for Multimodal Retrieval

We present ReMatch, a framework that leverages the generative strength of MLLMs for multimodal retrieval. Previous approaches treated an MLLM as a simple encoder, ignoring its generative nature, and under-utilising its compositional reasoning and world knowledge. We instead train the embedding MLLM end-to-end with a chat-style generative matching stage. The matching stage uses the same MLLM to autoregressively decide relevance from multi-view inputs, including both raw data and its own projected embeddings for each query and document. It provides instance-wise discrimination supervision that complements a standard contrastive loss, offering stronger gradients on hard negatives and preserving the compositional strengths of the original MLLM. To obtain semantically richer multimodal embeddings, we use multiple learnable tokens to augment each input, generating fine-grained contextual, mutually orthogonal embeddings with low inference cost. Leveraging our established high-performance baseline,we assemble the ideas mentioned above into a powerful training recipe and achieve a new state-of-the-art on the Massive Multimodal Embedding Benchmark (MMEB). Our experiments show particularly strong zero-shot generalization results on five datasets, highlighting the robustness and transferability of ReMatch.

cs.CV↗

arXiv2025

Doubly heavy hadron production in ultraperipheral collisions

The inclusive production of pseudoscalar heavy quarkonia ($η_c,\, η_b,\, B_c$), double heavy baryons $Ξ_{QQ^\prime}$ ($Q^{(\prime)}=c,\,b$ quarks) and tetraquarks $T_{QQ}$ in heavy ion ultraperipheral collisions (UPCs) is studied. Numerical results indicate that the experimental investigation of $η_c,\, Ξ_{cc}$, and $T_{cc}$ is feasible at the upcoming HL-LHC and future FCC. Heavy ion UPCs open another avenue for studying the production of these doubly heavy hadrons.

hep-ph↗

arXiv2025

Disentangling Causal Substructures for Interpretable and Generalizable Drug Synergy Prediction

Drug synergy prediction is a critical task in the development of effective combination therapies for complex diseases, including cancer. Although existing methods have shown promising results, they often operate as black-box predictors that rely predominantly on statistical correlations between drug characteristics and results. To address this limitation, we propose CausalDDS, a novel framework that disentangles drug molecules into causal and spurious substructures, utilizing the causal substructure representations for predicting drug synergy. By focusing on causal sub-structures, CausalDDS effectively mitigates the impact of redundant features introduced by spurious substructures, enhancing the accuracy and interpretability of the model. In addition, CausalDDS employs a conditional intervention mechanism, where interventions are conditioned on paired molecular structures, and introduces a novel optimization objective guided by the principles of sufficiency and independence. Extensive experiments demonstrate that our method outperforms baseline models, particularly in cold start and out-of-distribution settings. Besides, CausalDDS effectively identifies key substructures underlying drug synergy, providing clear insights into how drug combinations work at the molecular level. These results underscore the potential of CausalDDS as a practical tool for predicting drug synergy and facilitating drug discovery.

cs.LG↗