arXiv subjects

Jing Yang

Publications and source records attributed to Jing Yang.

At least 127 records · Page 7Linked to original sources

arXiv2025

Beam-commissioning-oriented optics study of HFRS Phase-I based on measured magnetic field data

The construction of the first phase of the High energy FRagment Separator (HFRS Phase-I) has already been completed and it is anticipated to start beam commissioning in autumn 2025. This paper presents the first order and higher order beam optics calculations for the HFRS Phase-I, using measured magnet data, and evaluates its experimental performance in preparation for beam commissioning. The first order optics of HFRS is calculated based on the sliced magnetic fields and the higher order aberrations are corrected using a self-compiled program. Monte Carlo particle tracking is employed to analyze the beam phase spaces on the focal planes. The experimental performance of the machine is evaluated through Monte Carlo simulations. The beam phase spaces on the focal planes are thoroughly examined, demonstrating that the higher order aberrations have been well corrected. Moreover, the experimental performance of HFRS is evaluated based on the corrected higher order optics, yielding satisfactory results: the secondary beams of interest can be well separated and exhibit high transmission efficiency. This work provides valuable insights for the upcoming beam commissioning of HFRS Phase-I. The effective correction of higher order aberrations and optimized magnet settings lay a solid foundation for future experiments.

physics.acc-ph↗

arXiv2025

AnyEnhance: A Unified Generative Model with Prompt-Guidance and Self-Critic for Voice Enhancement

We introduce AnyEnhance, a unified generative model for voice enhancement that processes both speech and singing voices. Based on a masked generative model, AnyEnhance is capable of handling both speech and singing voices, supporting a wide range of enhancement tasks including denoising, dereverberation, declipping, super-resolution, and target speaker extraction, all simultaneously and without fine-tuning. AnyEnhance introduces a prompt-guidance mechanism for in-context learning, which allows the model to natively accept a reference speaker's timbre. In this way, it could boost enhancement performance when a reference audio is available and enable the target speaker extraction task without altering the underlying architecture. Moreover, we also introduce a self-critic mechanism into the generative process for masked generative models, yielding higher-quality outputs through iterative self-assessment and refinement. Extensive experiments on various enhancement tasks demonstrate AnyEnhance outperforms existing methods in terms of both objective metrics and subjective listening tests. Demo audios are publicly available at https://amphionspace.github.io/anyenhance. An open-source implementation is provided at https://github.com/viewfinder-annn/anyenhance-v1-ccf-aatc.

cs.SD↗

arXiv2025

OracleAgent: A Multimodal Reasoning Agent for Oracle Bone Script Research

As one of the earliest writing systems, Oracle Bone Script (OBS) preserves the cultural and intellectual heritage of ancient civilizations. However, current OBS research faces two major challenges: (1) the interpretation of OBS involves a complex workflow comprising multiple serial and parallel sub-tasks, and (2) the efficiency of OBS information organization and retrieval remains a critical bottleneck, as scholars often spend substantial effort searching for, compiling, and managing relevant resources. To address these challenges, we present OracleAgent, the first agent system designed for the structured management and retrieval of OBS-related information. OracleAgent seamlessly integrates multiple OBS analysis tools, empowered by large language models (LLMs), and can flexibly orchestrate these components. Additionally, we construct a comprehensive domain-specific multimodal knowledge base for OBS, which is built through a rigorous multi-year process of data collection, cleaning, and expert annotation. The knowledge base comprises over 1.4M single-character rubbing images and 80K interpretation texts. OracleAgent leverages this resource through its multimodal tools to assist experts in retrieval tasks of character, document, interpretation text, and rubbing image. Extensive experiments demonstrate that OracleAgent achieves superior performance across a range of multimodal reasoning and generation tasks, surpassing leading mainstream multimodal large language models (MLLMs) (e.g., GPT-4o). Furthermore, our case study illustrates that OracleAgent can effectively assist domain experts, significantly reducing the time cost of OBS research. These results highlight OracleAgent as a significant step toward the practical deployment of OBS-assisted research and automated interpretation systems.

cs.CV↗

arXiv2025

Greedy Sampling Is Provably Efficient for RLHF

Reinforcement Learning from Human Feedback (RLHF) has emerged as a key technique for post-training large language models. Despite its empirical success, the theoretical understanding of RLHF is still limited, as learning the KL-regularized target with only preference feedback poses additional challenges compared with canonical RL. Existing works mostly study the reward-based Bradley-Terry (BT) preference model, and extend classical designs utilizing optimism or pessimism. This work, instead, considers the general preference model (whose practical relevance has been observed recently) and obtains performance guarantees with major, order-wise improvements over existing ones. Surprisingly, these results are derived from algorithms that directly use the empirical estimates (i.e., greedy sampling), as opposed to constructing optimistic or pessimistic estimates in previous works. This insight has a deep root in the unique structural property of the optimal policy class under the KL-regularized target, and we further specialize it to the BT model, highlighting the surprising sufficiency of greedy sampling in RLHF.

cs.LG↗

arXiv2025

User-defined Electrostatic Potentials in DFT Supercell Calculations: Implementation and Application to Electrified Interfaces

Introducing electric fields into density functional theory (DFT) calculations is essential for understanding electrochemical processes, interfacial phenomena, and the behavior of materials under applied bias. However, applying user-defined electrostatic potentials in DFT is nontrivial and often requires direct modification to the specific DFT code. In this work, we present an implementation for supercell DFT calculations under arbitrary electric fields and discuss the required corrections to the energies and forces. The implementation is realized through the recently released VASP-Python interface, enabling the application of user-defined fields directly within the standard VASP software and providing great flexibility and control. We demonstrate the application of this approach with diverse case studies, including molecular adsorption on electrified surfaces, field ion microscopy, electrochemical solid-water interfaces, and implicit solvent models.

cond-mat.mtrl-sci↗

arXiv2025

Backward-Friendly Optimization: Training Large Language Models with Approximate Gradients under Memory Constraints

Full fine-tuning of Large Language Models (LLMs) is notoriously memory-intensive, primarily because conventional optimizers such as SGD or Adam assume access to exact gradients derived from cached activations. Existing solutions either alter the model architecture (e.g., reversible networks) or trade memory for computation (e.g., activation checkpointing), but the optimizer itself remains untouched. In this work, we introduce GradLite, a backward-friendly optimizer that relaxes the requirement of exact gradients, enabling efficient training even when intermediate activations are aggressively discarded or approximated. GradLite leverages two key techniques: (i) low-rank Jacobian approximation, which reduces the dimensionality of backpropagated error signals, and (ii) error-feedback correction, which accumulates and compensates approximation errors across iterations to preserve convergence guarantees. We provide a theoretical analysis showing that GradLite maintains unbiased gradient estimates with bounded variance, ensuring convergence rates comparable to Adam. Empirically, GradLite reduces optimizer-state and activation memory consumption by up to 50\% without architectural changes, and achieves on-par or superior downstream performance on reasoning (MMLU, GSM8K), multilingual, and dialogue benchmarks compared to checkpointing and optimizer-centric baselines (LoMo, GaLore).

cs.LG↗

arXiv2025

DynaPose4D: High-Quality 4D Dynamic Content Generation via Pose Alignment Loss

Recent advancements in 2D and 3D generative models have expanded the capabilities of computer vision. However, generating high-quality 4D dynamic content from a single static image remains a significant challenge. Traditional methods have limitations in modeling temporal dependencies and accurately capturing dynamic geometry changes, especially when considering variations in camera perspective. To address this issue, we propose DynaPose4D, an innovative solution that integrates 4D Gaussian Splatting (4DGS) techniques with Category-Agnostic Pose Estimation (CAPE) technology. This framework uses 3D Gaussian Splatting to construct a 3D model from single images, then predicts multi-view pose keypoints based on one-shot support from a chosen view, leveraging supervisory signals to enhance motion consistency. Experimental results show that DynaPose4D achieves excellent coherence, consistency, and fluidity in dynamic motion generation. These findings not only validate the efficacy of the DynaPose4D framework but also indicate its potential applications in the domains of computer vision and animation production.

cs.CV↗

arXiv2025

Agent-GSPO: Communication-Efficient Multi-Agent Systems via Group Sequence Policy Optimization

To combat the prohibitive communication costs of ``free-for-all" multi-agent systems (MAS), we introduce \textbf{Agent-GSPO}, a framework that directly optimizes for token economy using sequence-level reinforcement learning. Agent-GSPO leverages the stable and memory-efficient Group Sequence Policy Optimization (GSPO) algorithm to train agents on a communication-aware reward that explicitly penalizes verbosity. Across seven reasoning benchmarks, Agent-GSPO not only achieves new state-of-the-art performance but does so with a fraction of the token consumption of existing methods. By fostering emergent strategies like ``strategic silence," our approach provides a practical blueprint for developing scalable and economically viable multi-agent systems.

cs.MA↗

arXiv2025

RaCoT: Plug-and-Play Contrastive Example Generation Mechanism for Enhanced LLM Reasoning Reliability

Retrieval-Augmented Generation (RAG) faces a core bottleneck with knowledge-sparse and semantically ambiguous long-tail queries, where retrieval noise distorts reasoning and necessitates costly post-processing. To tackle this, we propose RaCoT (Retrieval-aware Contrastive-of-Thought), a novel framework that shifts contrastive thinking to the pre-retrieval stage. By automatically generating a semantically adjacent yet differently answered contrastive question and extracting a $Δ$-Prompt to capture their key differences, RaCoT guides the model to proactively focus on the ``critical details that determine answer divergence." This approach allows it to suppress semantic interference within a single retrieval pass, overcoming the theoretical bottleneck of single-vector queries that struggle to simultaneously encode signals for what to attend to and what to ignore. On six authoritative benchmarks, including PopQA and TriviaQA-unfiltered, RaCoT outperforms strong baselines like RankRAG and Self-RAG by 0.9-2.4 percentage points. It exhibits superior robustness, with a performance drop of only 8.6\% in adversarial tests, far surpassing the over 15\% degradation in other methods. Furthermore, its low latency (3.12s) and token overhead (11.54) place it on the accuracy-efficiency Pareto frontier, while ablation studies validate the necessity of each component. Ultimately, RaCoT reframes the RAG paradigm from ``post-hoc context cleaning" to ``a priori shaping of discriminative reasoning", offering an efficient and robust path toward reliable AI systems for real-time, resource-constrained deployments.

cs.AI↗

arXiv2025

Guardian: Decoupling Exploration from Safety in Reinforcement Learning

Hybrid offline--online reinforcement learning (O2O RL) promises both sample efficiency and robust exploration, but suffers from instability due to distribution shift between offline and online data. We introduce RLPD-GX, a framework that decouples policy optimization from safety enforcement: a reward-seeking learner explores freely, while a projection-based guardian guarantees rule-consistent execution and safe value backups. This design preserves the exploratory value of online interactions without collapsing to conservative policies. To further stabilize training, we propose dynamic curricula that gradually extend temporal horizons and anneal offline--online data mixing. We prove convergence via a contraction property of the guarded Bellman operator, and empirically show state-of-the-art performance on Atari-100k, achieving a normalized mean score of 3.02 (+45\% over prior hybrid methods) with stronger safety and stability. Beyond Atari, ablations demonstrate consistent gains across safety-critical and long-horizon tasks, underscoring the generality of our design. Extensive and comprehensive results highlight decoupled safety enforcement as a simple yet principled route to robust O2O RL, suggesting a broader paradigm for reconciling exploration and safety in reinforcement learning.

cs.LG↗

arXiv2025

Position: The Artificial Intelligence and Machine Learning Community Should Adopt a More Transparent and Regulated Peer Review Process

The rapid growth of submissions to top-tier Artificial Intelligence (AI) and Machine Learning (ML) conferences has prompted many venues to transition from closed to open review platforms. Some have fully embraced open peer reviews, allowing public visibility throughout the process, while others adopt hybrid approaches, such as releasing reviews only after final decisions or keeping reviews private despite using open peer review systems. In this work, we analyze the strengths and limitations of these models, highlighting the growing community interest in transparent peer review. To support this discussion, we examine insights from Paper Copilot, a website launched two years ago to aggregate and analyze AI / ML conference data while engaging a global audience. The site has attracted over 200,000 early-career researchers, particularly those aged 18-34 from 177 countries, many of whom are actively engaged in the peer review process. Drawing on our findings, this position paper advocates for a more transparent, open, and well-regulated peer review aiming to foster greater community involvement and propel advancements in the field.

cs.DL↗

arXiv2025

Learning Dynamics of VLM Finetuning

Preference-based finetuning of vision--language models (VLMs) is brittle: trivially wrong negatives inject uninformative gradients that destabilize training. We recast alignment as \textbf{learning-dynamics--aware optimization} and introduce \textbf{Cooling-Weighted DPO (CW-DPO)}, a two-stage recipe that explicitly models and exploits the training trajectory. \textbf{Stage 1} performs supervised finetuning with \textbf{gentle negatives}: \textbf{low-weight smoothed supervision} that regularizes the base policy and curbs overconfidence without explicit penalties. \textbf{Stage 2} applies a DPO objective in which the \textbf{negative term is scaled by a cooling weight} computed from the model's \textbf{average token log-probability} on each negative, suppressing uninformative gradients from easy or off-distribution samples while preserving signal from hard negatives. In practice, we emphasize \textbf{on-policy negatives} and allow \textbf{mixed negatives} by blending a controllable fraction of dataset negatives to maintain contrast freshness. Throughout, we instrument training with $Δ\!\log p$ probes on positives and negatives as first-class signals for early stopping, curriculum design, and failure diagnosis. Across diverse VLM tasks, CW-DPO yields \textbf{more stable optimization}, \textbf{better calibration}, and \textbf{higher pairwise win-rates} than SFT-only and vanilla DPO, while \textbf{converging in fewer steps}. Ablations isolate the \textbf{cooling-weight mechanism} as the primary driver of these gains and show complementary benefits from mixing on-policy and dataset negatives. Taken together, our results show that \textbf{smoothing learning dynamics before cooling preferences} is a simple, general principle for robust VLM alignment.

cs.LG↗

arXiv2025

Lossless Postselected Quantum Metrology with Quasi-pure Mixed States

Postselection can compress the metrological information and improve sensitivity in the presence of certain types of technical noise. Postselected quantum metrology with pure states has been significantly advanced recently. However, extending this framework to mixed states leads to formidable challenges, such as the difficulty in searching for lossless postselection measurements or even the loss of metrological information. In this work, we leverage the intuition for the lossless postselection of pure states and generalize the theory to the lossless postselection of a class of mixed states, dubbed quasi-pure states. We illustrate our findings in postselected quantum imaging, unitary estimation problems, and show that the quasi-pure structure can be universally engineered through only classical correlation with an ancilla. Our findings extend the utility of postselection techniques to scenarios with decoherence and also offer new perspectives to foundational questions in quantum information geometry.

quant-ph↗

arXiv2025

The Astronomical Plate Digitization at SHAO

The digitization of historical astronomical plates is essential for preserving century-long observational data. This work presents the development and application of the specialized digitizers at the Shanghai Astronomical Observatory (SHAO), including technical details, international collaborations, and scientific applications on the plates.

astro-ph.IM↗

arXiv2025

Macro-Dipole-Constrainted Learning of Atomic Charges for Accurate Electrostatic Potentials at Electrochemical Interfaces

Large thermal fluctuations of the liquid phase obscure the weak macroscopic electric field that drives electrochemical reactions, rendering the extraction of reliable interfacial charge distributions from ab initio molecular dynamics extremely challenging. We introduce SMILE-CP (Scalar Macro-dipole Integrated LEarning - Charge Partitioning), a macro-dipole-constrained scheme that infers atomic charges using only the instantaneous atomic coordinates and the total dipole moment of the simulation cell - quantities routinely available from standard density-functional theory calculations. SMILE-CP preserves both the global electrostatic field and the local potential without invoking any explicit charge-partitioning scheme. Benchmarks on three representative electrochemical interfaces - nanoconfined water, Mg2+ dissolution in water, and a kinked Mg vicinal surface under anodic bias - show that SMILE-CP eliminates the qualitative errors observed for unconstrained charge decompositions. The method is computationally inexpensive and data-efficient, opening the door to charge-aware machine-learning potentials capable of bias-controlled, nanosecond-scale simulations of realistic electrochemical systems.

physics.chem-ph↗

arXiv2025

Stratifying Discriminant Hypersurface

This paper investigates the stratification of the discriminant hypersurface associated with a univariate polynomial via the number of its distinct complex roots. We introduce two novel approaches different from the one based on subdiscriminants. The first approach stratifies the discriminant hypersurface by recursively removing all the lowest-order points, while the second one stratifies the discriminant hypersurface by recursively removing all the smooth points. Both approaches rely solely on the discriminant itself instead of using high-order subdiscriminants. These results offer new insights into the intrinsic geometry of the discriminant and its connection to root multiplicity.

math.AG↗

arXiv2025

SingVERSE: A Diverse, Real-World Benchmark for Singing Voice Enhancement

This paper presents a benchmark for singing voice enhancement. The development of singing voice enhancement is limited by the lack of realistic evaluation data. To address this gap, this paper introduces SingVERSE, the first real-world benchmark for singing voice enhancement, covering diverse acoustic scenarios and providing paired, studio-quality clean references. Leveraging SingVERSE, we conduct a comprehensive evaluation of state-of-the-art models and uncover a consistent trade-off between perceptual quality and intelligibility. Finally, we show that training on in-domain singing data substantially improves enhancement performance without degrading speech capabilities, establishing a simple yet effective path forward. This work offers the community a foundational benchmark together with critical insights to guide future advances in this underexplored domain. Demopage: https://singverse.github.io

cs.SD↗

arXiv2025

Evaluating Cognitive-Behavioral Fixation via Multimodal User Viewing Patterns on Social Media

Digital social media platforms frequently contribute to cognitive-behavioral fixation, a phenomenon in which users exhibit sustained and repetitive engagement with narrow content domains. While cognitive-behavioral fixation has been extensively studied in psychology, methods for computationally detecting and evaluating such fixation remain underexplored. To address this gap, we propose a novel framework for assessing cognitive-behavioral fixation by analyzing users' multimodal social media engagement patterns. Specifically, we introduce a multimodal topic extraction module and a cognitive-behavioral fixation quantification module that collaboratively enable adaptive, hierarchical, and interpretable assessment of user behavior. Experiments on existing benchmarks and a newly curated multimodal dataset demonstrate the effectiveness of our approach, laying the groundwork for scalable computational analysis of cognitive fixation. All code in this project is publicly available for research purposes at https://github.com/Liskie/cognitive-fixation-evaluation.

cs.SI↗