arXiv subjects

Gang Li

Publications and source records attributed to Gang Li.

At least 127 records · Page 7Linked to original sources

arXiv2026

Parameter Convergence Radar Detector Based on VAMP Deep Unfolding

Compared with the sparse recovery process in traditional compressed sensing (CS) radar detector CAMP, vector AMP deep unfolding (VAMP-DU) can achieve sparse recovery over a broader range of observation matrices, with faster convergence speed and higher recovery accuracy. However, the distribution of the error term in VAMP-DU remains unknown, which renders the distribution of the test statistic in CS radar detection undetermined and thus hinders threshold setting under a given false alarm rate when VAMP-DU is applied to CS radar detection. In this work, we theoretically prove that the error term in VAMP-DU follows a Gaussian distribution by leveraging a general state evolution (SE). Based on the Gaussianity, we propose a new parameter convergence radar detector (PCRD) as the CS detector to calculate the distribution parameter of the test statistic and realize target detection under a given false alarm rate. Specifically, PCRD exploits the Gaussian property of error term in VAMP-DU to exhibit superior false alarm control capability, while leveraging the improved recovery accuracy of VAMP-DU to further enhance target detection performance. Numerical simulations validate the Gaussianity of the error term in VAMP-DU and show the superiority of the VAMP-DU-based PCRD over existing approaches in both false alarm control accuracy and target detection performance.

eess.SP↗

arXiv2026

DisCO: Reinforcing Large Reasoning Models with Discriminative Constrained Optimization

The recent success and openness of DeepSeek-R1 have brought widespread attention to Group Relative Policy Optimization (GRPO) as a reinforcement learning method for large reasoning models (LRMs). In this work, we analyze the GRPO objective under a binary reward setting and reveal an inherent limitation of question-level difficulty bias. We also identify a connection between GRPO and traditional discriminative methods in supervised learning. Motivated by these insights, we introduce a new Discriminative Constrained Optimization (DisCO) framework for reinforcing LRMs, grounded in the principle of discriminative learning. The main differences between DisCO and GRPO and its recent variants are: (1) it replaces the group relative objective with a discriminative objective defined by a scoring function; (2) it abandons clipping-based surrogates in favor of non-clipping RL surrogate objectives used as scoring functions; (3) it employs a simple yet effective constrained optimization approach to enforce the KL divergence constraint. As a result, DisCO offers notable advantages over GRPO and its variants: (i) it completely eliminates difficulty bias by adopting discriminative objectives; (ii) it addresses the entropy instability in GRPO and its variants through the use of non-clipping scoring functions and a constrained optimization approach, yielding long and stable training dynamics; (iii) it allows the incorporation of advanced discriminative learning techniques to address data imbalance, where a significant number of questions have more negative than positive generated answers during training. Our experiments on enhancing the mathematical reasoning capabilities of SFT-finetuned models show that DisCO significantly outperforms GRPO and its improved variants such as DAPO, achieving average gains of 7\% over GRPO and 6\% over DAPO across six benchmark tasks for a 1.5B model.

cs.LG↗

arXiv2026

SmartSnap: Proactive Evidence Seeking for Self-Verifying Agents

Agentic reinforcement learning (RL) holds great promise for the development of autonomous agents under complex GUI tasks, but its scalability remains severely hampered by the verification of task completion. Existing task verification is treated as a passive, post-hoc process: a verifier (i.e., rule-based scoring script, reward or critic model, and LLM-as-a-Judge) analyzes the agent's entire interaction trajectory to determine if the agent succeeds. Such processing of verbose context that contains irrelevant, noisy history poses challenges to the verification protocols and therefore leads to prohibitive cost and low reliability. To overcome this bottleneck, we propose SmartSnap, a paradigm shift from this passive, post-hoc verification to proactive, in-situ self-verification by the agent itself. We introduce the Self-Verifying Agent, a new type of agent designed with dual missions: to not only complete a task but also to prove its accomplishment with curated snapshot evidences. Guided by our proposed 3C Principles (Completeness, Conciseness, and Creativity), the agent leverages its accessibility to the online environment to perform self-verification on a minimal, decisive set of snapshots. Such evidences are provided as the sole materials for a general LLM-as-a-Judge verifier to determine their validity and relevance. Experiments on mobile tasks across model families and scales demonstrate that our SmartSnap paradigm allows training LLM-driven agents in a scalable manner, bringing performance gains up to 26.08% and 16.66% respectively to 8B and 30B models. The synergizing between solution finding and evidence seeking facilitates the cultivation of efficient, self-verifying agents with competitive performance against DeepSeek V3.1 and Qwen3-235B-A22B. Code is available at: https://github.com/TencentYoutuResearch/SmartSnap

cs.CL↗

arXiv2026

Charge disproportionation as a possible mechanism towards polar antiferromagnetic metal in molecular orbital crystal

Polar antiferromagnetic metals have recently garnered increasing interests due to their combined traits of both ferromagnets and antiferromagnets for spintronic applications. However, the inherently incompatible nature of antiferromagnet, metallicity and polarity pose a significant challenge. We propose that charge disproportionation can lead to this novel state in negative charge transfer gap regime in molecular orbital crystal by molecular orbital analyses of first-principles DFT+$U$ electronic band structure for representative Ruddlesden-Popper bilayer perovskite oxides Sr$_3$Co$_2$O$_7$, corroborated by Density Matrix Renormalization Group calculation. Due to the negative charge transfer nature of Co$^{4+}$ and imposed by strong interlayer coupling, localized molecular orbitals stemming from the hybridization of Co $d_{z^2}$ and $d_{xz/yz}$ orbitals through the apical oxygen $p$ orbitals are preferably emergent within each bilayer unit, which develop antiferromagnetic ordering by invoking Hubbard repulsion. Charge disproportionation driven by Hund's physics, makes an occupation imbalance with broken inversion symmetry in the remaining $d_{xy}$ and $d_{x^2-y^2}$ orbitals from distinct Co atoms within the bilayer unit, resulting in the polar metallicity. Meanwhile, this charge disproportionation scenario allows consequent conducting carriers to couple with interlayer local spins via Hund's coupling, giving rise to in-plane double-exchange ferromagnetism. Our molecular orbital formulation further provides a guide towards an effective Hamiltonian for modelling the unconventional synergy of metallicity, polarity and antiferromagnetism in Sr$_3$Co$_2$O$_7$, which may be a unified framework widely applicable to double-layer Ruddlesden-Popper perovskite oxides.

cond-mat.mtrl-sci↗

arXiv2025

Youtu-Agent: Scaling Agent Productivity with Automated Generation and Hybrid Policy Optimization

Existing Large Language Model (LLM) agent frameworks face two significant challenges: high configuration costs and static capabilities. Building a high-quality agent often requires extensive manual effort in tool integration and prompt engineering, while deployed agents struggle to adapt to dynamic environments without expensive fine-tuning. To address these issues, we propose \textbf{Youtu-Agent}, a modular framework designed for the automated generation and continuous evolution of LLM agents. Youtu-Agent features a structured configuration system that decouples execution environments, toolkits, and context management, enabling flexible reuse and automated synthesis. We introduce two generation paradigms: a \textbf{Workflow} mode for standard tasks and a \textbf{Meta-Agent} mode for complex, non-standard requirements, capable of automatically generating tool code, prompts, and configurations. Furthermore, Youtu-Agent establishes a hybrid policy optimization system: (1) an \textbf{Agent Practice} module that enables agents to accumulate experience and improve performance through in-context optimization without parameter updates; and (2) an \textbf{Agent RL} module that integrates with distributed training frameworks to enable scalable and stable reinforcement learning of any Youtu-Agents in an end-to-end, large-scale manner. Experiments demonstrate that Youtu-Agent achieves state-of-the-art performance on WebWalkerQA (71.47\%) and GAIA (72.8\%) using open-weight models. Our automated generation pipeline achieves over 81\% tool synthesis success rate, while the Practice module improves performance on AIME 2024/2025 by +2.7\% and +5.4\% respectively. Moreover, our Agent RL training achieves 40\% speedup with steady performance improvement on 7B LLMs, enhancing coding/reasoning and searching capabilities respectively up to 35\% and 21\% on Maths and general/multi-hop QA benchmarks.

cs.AI↗

arXiv2025

Dissecting Lepton Number Violating Interactions in the Left-Right Symmetric Model: $0νββ$ decay, Møller scattering, and collider searches

In the context of the left-right symmetric model, we study the interplay of neutrinoless double beta ($0νββ$) decay, parity-violating Møller scattering, and high-energy colliders, resulting from the Yukawa interaction of the right-handed doubly-charged scalar to electrons, which could evade the severe constraints from charged lepton flavor violation. The $0νββ$ decay amplitude receives additional contributions from right-handed sterile neutrinos. The half-life, calculated in the effective field theory (EFT) framework, allows for an improved description of the contributions involving non-zero mixing between left- and right-handed $W$ bosons and those arising from exchanging a light right-handed neutrino. We find that the relative sensitivities between the low-energy (or high-precision) and high-energy experiments are affected by the left-right mixing. On the other hand, our results show how the interplay of collider and low-energy searches provides a manner to explore regions that are inaccessible to $0νββ$ decay experiments.

hep-ph↗

arXiv2025

Multi Agents Semantic Emotion Aligned Music to Image Generation with Music Derived Captions

When people listen to music, they often experience rich visual imagery. We aim to externalize this inner imagery by generating images conditioned on music. We propose MESA MIG, a multi agent semantic and emotion aligned framework that first produces structured music captions and then refines them with cooperating agents specializing in scene, motion, style, color, and composition. In parallel, a Valence Arousal regression head predicts continuous affective states from music, while a CLIP based visual VA head estimates emotions from images. These components jointly enforce semantic and emotional alignment between music and synthesized images. Experiments on curated music image pairs show that MESA MIG outperforms caption only and single agent baselines in aesthetic quality, semantic consistency, and VA alignment, and achieves competitive emotion regression performance compared with state of the art music and image emotion models.

cs.MM↗

arXiv2025

Structured Event Representation and Stock Return Predictability

We find that event features extracted by large language models (LLMs) are effective for text-based stock return prediction. Using a pre-trained LLM to extract event features from news articles, we propose a novel deep learning model based on structured event representation (SER) and attention mechanisms to predict stock returns in the cross-section. Our SER-based model provides superior performance compared with other existing text-driven models to forecast stock returns out of sample and offers highly interpretable feature structures to examine the mechanisms underlying the stock return predictability. We further provide various implications based on SER and highlight the crucial benefit of structured model inputs in stock return predictability.

econ.GN↗

arXiv2025

Kinetics of Bose-Einstein condensation of magnons in Yttrium Iron Garnet films

In this article, we explain the reason of the apparent contradiction between recent experiments [1] and [2] and earlier theoretical predictions [3] of strongly asymmetric condensate resulting in attractive interaction between the condensate magnons. We show that the relaxation time for equilibrium between two condensates at two minima of energy exceeds the time of experiment. Therefore, it should be described by Boltzmann kinetic equation. We develop the proper kinetic theory and find the relation between the critical pumping power and the effective temperature of over-condensate magnons.

cond-mat.mes-hall↗

arXiv2025

Black-Box Auditing of Quantum Model: Lifted Differential Privacy with Quantum Canaries

Quantum machine learning (QML) promises significant computational advantages, yet models trained on sensitive data risk memorizing individual records, creating serious privacy vulnerabilities. While Quantum Differential Privacy (QDP) mechanisms provide theoretical worst-case guarantees, they critically lack empirical verification tools for deployed models. We introduce the first black-box privacy auditing framework for QML based on Lifted Quantum Differential Privacy, leveraging quantum canaries (strategically offset-encoded quantum states) to detect memorization and precisely quantify privacy leakage during training. Our framework establishes a rigorous mathematical connection between canary offset and trace distance bounds, deriving empirical lower bounds on privacy budget consumption that bridge the critical gap between theoretical guarantees and practical privacy verification. Comprehensive evaluations across both simulated and physical quantum hardware demonstrate our framework's effectiveness in measuring actual privacy loss in QML models, enabling robust privacy verification in QML systems.

cs.LG↗

arXiv2025

Citation importance-aware document representation learning for large-scale science mapping

Effective science mapping relies on high-quality representations of scientific documents. As an important task in scientometrics and information studies, science mapping is often challenged by the complex and heterogeneous nature of citations. While previous studies have attempted to improve document representations by integrating citation and semantic information, the heterogeneity of citations is often overlooked. To address this problem, this study proposes a citation importance-aware contrastive learning framework that refines the supervisory signal. We first develop a scalable measurement of citation importance based on location, frequency, and self-citation characteristics. Citation importance is then integrated into the contrastive learning process through an importance-aware sampling strategy, which selects low-importance citations as hard negatives. This forces the model to learn finer-grained representations that distinguish between important and perfunctory citations. To validate the effectiveness of the proposed framework, we fine-tune a SciBERT model and perform extensive evaluations on SciDocs and PubMed benchmark datasets. Results show consistent improvements in both document representation quality and science mapping accuracy. Furthermore, we apply the trained model to over 33 million documents from Web of Science. The resulting map of science accurately visualizes the global and local intellectual structure of science and reveals interdisciplinary research fronts. By operationalizing citation heterogeneity into a scalable computational framework, this study demonstrates how differentiating citations by their importance can be effectively leveraged to improve document representation and science mapping.

cs.DL↗

arXiv2025

Learn the Ropes, Then Trust the Wins: Self-imitation with Progressive Exploration for Agentic Reinforcement Learning

Reinforcement learning (RL) is the dominant paradigm for sharpening strategic tool use capabilities of LLMs on long-horizon, sparsely-rewarded agent tasks, yet it faces a fundamental challenge of exploration-exploitation trade-off. Existing studies stimulate exploration through the lens of policy entropy, but such mechanical entropy maximization is prone to RL instability due to the multi-turn distribution shifting. In this paper, we target the progressive exploration-exploitation balance under the guidance of the agent's own experiences without succumbing to either entropy collapsing or runaway divergence. We propose SPEAR, a self-imitation learning (SIL) recipe for training agentic LLMs. It extends the vanilla SIL, where a replay buffer stores good experience for off-policy update, by gradually steering the policy entropy across stages. Specifically, the proposed curriculum scheduling harmonizes intrinsic reward shaping and self-imitation to 1) expedite exploration via frequent tool interactions at the beginning, and 2) strengthen exploitation of successful tactics upon convergence towards familiarity with the environment. We also combine bag-of-tricks of industrial RL optimizations for a strong baseline Dr.BoT to demonstrate our effectiveness. In ALFWorld and WebShop, SPEAR increases the success rates of GRPO/GiGPO/Dr.BoT by up to 16.1%/5.1%/8.6% and 20.7%/11.8%/13.9%, respectively. In AIME24 and AIME25, SPEAR boosts Dr.BoT by up to 3.8% and 6.1%, respectively. Such gains incur only 10%-25% extra theoretical complexity and negligible runtime overhead in practice, demonstrating the plug-and-play scalability of SPEAR.

cs.LG↗

arXiv2025

Towards Efficient and Accurate Spiking Neural Networks via Adaptive Bit Allocation

Multi-bit spiking neural networks (SNNs) have recently become a heated research spot, pursuing energy-efficient and high-accurate AI. However, with more bits involved, the associated memory and computation demands escalate to the point where the performance improvements become disproportionate. Based on the insight that different layers demonstrate different importance and extra bits could be wasted and interfering, this paper presents an adaptive bit allocation strategy for direct-trained SNNs, achieving fine-grained layer-wise allocation of memory and computation resources. Thus, SNN's efficiency and accuracy can be improved. Specifically, we parametrize the temporal lengths and the bit widths of weights and spikes, and make them learnable and controllable through gradients. To address the challenges caused by changeable bit widths and temporal lengths, we propose the refined spiking neuron, which can handle different temporal lengths, enable the derivation of gradients for temporal lengths, and suit spike quantization better. In addition, we theoretically formulate the step-size mismatch problem of learnable bit widths, which may incur severe quantization errors to SNN, and accordingly propose the step-size renewal mechanism to alleviate this issue. Experiments on various datasets, including the static CIFAR and ImageNet datasets and the dynamic CIFAR-DVS and DVS-GESTURE datasets, demonstrate that our methods can reduce the overall memory and computation cost while achieving higher accuracy. Particularly, our SEWResNet-34 can achieve a 2.69\% accuracy gain and 4.16$\times$ lower bit budgets over the advanced baseline work on ImageNet. This work is open-sourced at \href{https://github.com/Ikarosy/Towards-Efficient-and-Accurate-Spiking-Neural-Networks-via-Adaptive-Bit-Allocation}{this link}.

cs.NE↗

arXiv2025

FedCure: Mitigating Participation Bias in Semi-Asynchronous Federated Learning with Non-IID Data

While semi-asynchronous federated learning (SAFL) combines the efficiency of synchronous training with the flexibility of asynchronous updates, it inherently suffers from participation bias, which is further exacerbated by non-IID data distributions. More importantly, hierarchical architecture shifts participation from individual clients to client groups, thereby further intensifying this issue. Despite notable advancements in SAFL research, most existing works still focus on conventional cloud-end architectures while largely overlooking the critical impact of non-IID data on scheduling across the cloud-edge-client hierarchy. To tackle these challenges, we propose FedCure, an innovative semi-asynchronous Federated learning framework that leverages coalition construction and participation-aware scheduling to mitigate participation bias with non-IID data. Specifically, FedCure operates through three key rules: (1) a preference rule that optimizes coalition formation by maximizing collective benefits and establishing theoretically stable partitions to reduce non-IID-induced performance degradation; (2) a scheduling rule that integrates the virtual queue technique with Bayesian-estimated coalition dynamics, mitigating efficiency loss while ensuring mean rate stability; and (3) a resource allocation rule that enhances computational efficiency by optimizing client CPU frequencies based on estimated coalition dynamics while satisfying delay requirements. Comprehensive experiments on four real-world datasets demonstrate that FedCure improves accuracy by up to 5.1x compared with four state-of-the-art baselines, while significantly enhancing efficiency with the lowest coefficient of variation 0.0223 for per-round latency and maintaining long-term balance across diverse scenarios.

cs.LG↗

arXiv2025

ChemFM as a Scaling Law Guided Foundation Model Pre-trained on Informative Chemicals

Traditional AI methods often rely on task-specific model designs and training, which constrain both the scalability of model size and generalization across different tasks. Here, we introduce ChemFM, a large foundation model specifically developed for chemicals. By conducting a series of scaling experiments, we identify UniChem as the informative molecular database for pre-training the foundation model. ChemFM comprises 3 billion parameters and is pre-trained on 178 million molecules using self-supervised causal language modeling to extract generalizable molecular representations. This model can be adapted to diverse downstream chemical applications using either full-parameter or parameter-efficient fine-tuning methods. ChemFM consistently outperforms state-of-the-art task-specific AI models across all tested tasks. Notably, it achieves up to 67.48% performance improvement across 34 property prediction benchmarks, up to 33.80% reduction in mean average deviation between conditioned and actual properties of generated molecules in conditional molecular generation tasks, and up to 3.7% top-1 accuracy improvement across 4 reaction prediction datasets. Moreover, ChemFM demonstrates its superior performance in predicting antibiotic activity and cytotoxicity, highlighting its potential to advance the discovery of novel antibiotics. Furthermore, we demonstrate that, as a foundation model, ChemFM exhibits strong data efficiency, requiring significantly fewer labeled training samples to achieve state-of-the-art performance. We anticipate that ChemFM will significantly advance chemistry research by providing a foundation model capable of effectively generalizing across a broad range of tasks with minimal additional training.

cs.CE↗

arXiv2025

FLYINGTRUST: A Benchmark for Quadrotor Navigation Across Scenarios and Vehicles

Visual navigation algorithms for quadrotors often exhibit a large variation in performance when transferred across different vehicle platforms and scene geometries, which increases the cost and risk of field deployment. To support systematic early-stage evaluation, we introduce FLYINGTRUST, a high-fidelity, configurable benchmarking framework that measures how platform kinodynamics and scenario structure jointly affect navigation robustness. FLYINGTRUST models vehicle capability with two compact, physically interpretable indicators: maximum thrust-to-weight ratio and axis-wise maximum angular acceleration. The benchmark pairs a diverse scenario library with a heterogeneous set of real and virtual platforms and prescribes a standardized evaluation protocol together with a composite scoring method that balances scenario importance, platform importance and performance stability. We use FLYINGTRUST to compare representative optimization-based and learning-based navigation approaches under identical conditions, performing repeated trials per platform-scenario combination and reporting uncertainty-aware metrics. The results reveal systematic patterns: navigation success depends predictably on platform capability and scene geometry, and different algorithms exhibit distinct preferences and failure modes across the evaluated conditions. These observations highlight the practical necessity of incorporating both platform capability and scenario structure into algorithm design, evaluation, and selection, and they motivate future work on methods that remain robust across diverse platforms and scenarios.

cs.RO↗

arXiv2025

CUARewardBench: A Benchmark for Evaluating Reward Models on Computer-using Agent

Computer-using agents (CUAs) enable task completion through natural interaction with operating systems and software interfaces. While script-based verifiers are widely adopted for evaluation, they suffer from limited scalability and inability to provide step-wise assessment. Reward models offer promising alternatives, but their effectiveness on CUA evaluation remains largely underexplored. To address this gap, we present CUARewardBench, comprising four key contributions: (1) First-ever Comprehensive CUA Reward Benchmark: We introduce the first benchmark for evaluating both outcome reward models (ORM) and process reward models (PRM) on CUA tasks, enabling systematic assessment across trajectory-level and step-level evaluation. (2) Diverse, Practical and Reliable Dataset: CUARewardBench encompasses trajectories from 10 software categories and 7 agent architectures with varying performance levels (25.9%-50.8% success rates). All trajectories are expertly annotated through carefully designed protocols, with rigorous quality control to ensure reliability and practical applicability. (3) Comprehensive Analysis and Insights: Through extensive experiments across 7 vision-language models and 3 prompt templates, we reveal critical limitations of current CUA RMs, including insufficient visual reasoning capabilities, knowledge deficiencies, and the superiority of general VLMs over specialized CUA models for reward evaluation. (4) Unanimous Prompt Ensemble (UPE): Based on the insights from our comprehensive analysis, we propose UPE, a novel ensemble method that significantly enhances reward model reliability through strict unanimous voting and strategic prompt-template configurations. UPE achieves 89.8% precision and 93.3% NPV for ORM, and 81.7% precision and 85.1% NPV for PRM, substantially outperforming single VLMs and traditional ensemble approaches.

cs.SE↗

arXiv2025

Dipionic transitions of $Y(4500)$ to $J/ψ$

We, using an effective Lagrangian approach, investigated the dipionic transition of the $Y(4500)$ newly well established in the process $e^+e^-\to K^+K^- J/ψ$. In this study, the $Y(4500)$ was considered as a mixture of the $ψ(5S)$ and $ψ(4\,{^3D_1})$, and its dipionic decay to the $J/ψ$ was assumed to occur via charmed meson loops including box and two kinds of triangle loops. The calculated invariant mass spectra due to the different loop mechanisms exhibit distinct differences. These spectrum patterns are found to be nearly independent of the cutoff parameter, especially for the box and $\mathcal{R}$ loops. By means of comparing our calculated results with the future experimental measurements of the $π^+π^-$ and $π^\pm J/ψ$ invariant mass distributions, we could judge which loop mechanism is of more importance in this dipionic transition. The interference among the different kinds of loops is also exhibited. Despite the unknown phase angles between different kinds of loop diagrams, the partial decay width for the process $Y(4500)\toπ^+π^- J/ψ$ is estimated to be 250--700 keV, aligning with the estimation obtained by combining BESIII Collaboration measurements and theoretical predictions for $Y(4500)\to K^+K^- J/ψ$. We hope that the present calculations would be tested by the future BESIII or Belle II experiments.

hep-ph↗