arXiv subjects

Chao Wang

Publications and source records attributed to Chao Wang.

At least 181 records · Page 10Linked to original sources

arXiv2026

FinDeepResearch: Evaluating Deep Research Agents in Rigorous Financial Analysis

Deep Research (DR) agents, powered by advanced Large Language Models (LLMs), have recently garnered increasing attention for their capability in conducting complex research tasks. However, existing literature lacks a rigorous and systematic evaluation of DR Agent's capabilities in critical research analysis. To address this gap, we first propose HisRubric, a novel evaluation framework with a hierarchical analytical structure and a fine-grained grading rubric for rigorously assessing DR agents' capabilities in corporate financial analysis. This framework mirrors the professional analyst's workflow, progressing from data recognition to metric calculation, and finally to strategic summarization and interpretation. Built on this framework, we construct a FinDeepResearch benchmark that comprises 64 listed companies from 8 financial markets across 4 languages, encompassing a total of 15,808 grading items. We further conduct extensive experiments on the FinDeepResearch using 16 representative methods, including 6 DR agents, 5 LLMs equipped with both deep reasoning and search capabilities, and 5 LLMs with deep reasoning capabilities only. The results reveal the strengths and limitations of these approaches across diverse capabilities, financial markets, and languages, offering valuable insights for future research and development. The benchmark and evaluation code is publicly available at https://OpenFinArena.com/.

cs.CL↗

arXiv2026

Crystal-KV: Efficient KV Cache Management for Chain-of-Thought LLMs via Answer-First Principle

Chain-of-Thought (CoT) reasoning in large language models (LLMs) significantly improves accuracy on complex tasks, yet incurs excessive memory overhead due to the long think-stage sequences stored in the Key-Value (KV) cache. Unlike traditional generation tasks where all tokens are uniformly important, CoT emphasizes the final answer, rendering conventional KV compression strategies ineffective. In this paper, we present Crystal-KV, an efficient KV cache management framework tailored for CoT reasoning. Our key insight is the answer-first principle. By mapping answer preferences into think-stage attention map, we distinguish between SlipKV, which mainly maintains the reasoning flow but may occasionally introduce misleading context, and CrystalKV, which truly contributes to the correctness of the final answer. Next, we propose an attention-based Least Recently Frequently Used algorithm. It precisely identifies when a SlipKV entry's utility expires and evicts it, retaining CrystalKV without disrupting reasoning flow. Finally, we introduce an adaptive cache budget allocation algorithm. Based on the dynamic proportion of CrystalKV, it estimates the importance of each layer/head and adjusts the KV cache budget during inference, amplifying critical components to improve budget utilization. Results show that Crystal-KV achieves state-of-the-art KV cache compression, significantly improves throughput, and enables faster response time, while maintaining, or even improving, answer accuracy for CoT reasoning.

cs.CL↗

arXiv2025

xVerify: Efficient Answer Verifier for Reasoning Model Evaluations

With the release of OpenAI's o1 model, reasoning models that adopt slow-thinking strategies have become increasingly common. Their outputs often contain complex reasoning, intermediate steps, and self-reflection, making existing evaluation methods and reward models inadequate. In particular, they struggle to judge answer equivalence and to reliably extract final answers from long, complex responses. To address this challenge, we propose xVerify, an efficient answer verifier for evaluating reasoning models. xVerify shows strong equivalence judgment capabilities, enabling accurate comparison between model outputs and reference answers across diverse question types. To train and evaluate xVerify, we construct the VAR dataset, which consists of question-answer pairs generated by multiple LLMs across various datasets. The dataset incorporates multiple reasoning models and challenging evaluation sets specifically designed for reasoning assessment, with a multi-round annotation process to ensure label quality. Based on VAR, we train xVerify models at different scales. Experimental results on both test and generalization sets show that all xVerify variants achieve over 95% F1 score and accuracy. Notably, the smallest model, xVerify-0.5B-I, outperforms all evaluation methods except GPT-4o, while xVerify-3B-Ib surpasses GPT-4o in overall performance. In addition, reinforcement learning experiments using xVerify as the reward model yield an 18.4% improvement for Qwen2.5-7B compared with direct generation, exceeding the gains achieved with Math Verify as the reward. These results demonstrate the effectiveness and generalizability of xVerify. All xVerify resources are available on \href{https://github.com/IAAR-Shanghai/xVerify}{GitHub}.

cs.CL↗

arXiv2025

Training Report of TeleChat3-MoE

TeleChat3-MoE is the latest series of TeleChat large language models, featuring a Mixture-of-Experts (MoE) architecture with parameter counts ranging from 105 billion to over one trillion,trained end-to-end on Ascend NPU cluster. This technical report mainly presents the underlying training infrastructure that enables reliable and efficient scaling to frontier model sizes. We detail systematic methodologies for operator-level and end-to-end numerical accuracy verification, ensuring consistency across hardware platforms and distributed parallelism strategies. Furthermore, we introduce a suite of performance optimizations, including interleaved pipeline scheduling, attention-aware data scheduling for long-sequence training,hierarchical and overlapped communication for expert parallelism, and DVM-based operator fusion. A systematic parallelization framework, leveraging analytical estimation and integer linear programming, is also proposed to optimize multi-dimensional parallelism configurations. Additionally, we present methodological approaches to cluster-level optimizations, addressing host- and device-bound bottlenecks during large-scale training tasks. These infrastructure advancements yield significant throughput improvements and near-linear scaling on clusters comprising thousands of devices, providing a robust foundation for large-scale language model development on hardware ecosystems.

cs.CL↗

arXiv2025

Transition threshold for the Navier-Stokes-Coriolis system at high Reynolds numbers

The transition mechanism from laminar flow to turbulent flow is a central problem in hydrodynamic stability theory. To shed light on this transition mechanism, Trefethen et al.({\it \small Science 1993}) proposed the transition threshold problem, aiming to quantify the magnitude of perturbations required to trigger instability and determine their scaling with the Reynolds number. In this paper, we investigate the transition threshold of Couette flow for the three-dimensional incompressible Navier-Stokes-Coriolis system in the high Reynolds number regime ($\mathrm{Re}\gg 1$). By exploiting the combined effects of rotation (dispersion) and mixing mechanisms, we derive an improved stability threshold scaling in $\mathrm{Re}$. Precisely, we show that if the initial perturbation satisfies $$\|v_{in}-(y, 0, 0)\|_{\tilde{H}(\mathbb T \times \mathbb D)}\leq ε_0 \,\mathrm{Re}^{-α},$$ with any $α>\frac 23$ and $\tilde{H}=H^6 \cap W^{3,1}$ for $\mathbb D=\mathbb{R}^2$, and with any $α\geq\frac 56$ and $\tilde{H}=H^6$ for $\mathbb D=\mathbb{R}\times\mathbb{T}$, the corresponding solution of the Navier-Stokes-Coriolis system exists globally in time and remains asymptotically close to the Couette flow. The main analytical challenge arises from the anisotropic nature of the estimates for the zero modes and from the interactions between zero and non-zero modes, which we address using an anisotropic Sobolev space directly tailored to the zero modes. Additionally, we introduce a new dispersive structure for the zero modes and derive suitable Strichartz-type estimates. These tools enable us to exploit both the nonlinear structure and the improved dispersive behavior of certain good components of the zero modes, which play a crucial role in achieving the improved stability threshold.

math.AP↗

arXiv2025

A Large Language Model Based Method for Complex Logical Reasoning over Knowledge Graphs

Reasoning over knowledge graphs (KGs) with first-order logic (FOL) queries is challenging due to the inherent incompleteness of real-world KGs and the compositional complexity of logical query structures. Most existing methods rely on embedding entities and relations into continuous geometric spaces and answer queries via differentiable set operations. While effective for simple query patterns, these approaches often struggle to generalize to complex queries involving multiple operators, deeper reasoning chains, or heterogeneous KG schemas. We propose ROG (Reasoning Over knowledge Graphs with large language models), an ensemble-style framework that combines query-aware KG neighborhood retrieval with large language model (LLM)-based chain-of-thought reasoning. ROG decomposes complex FOL queries into sequences of simpler sub-queries, retrieves compact, query-relevant subgraphs as contextual evidence, and performs step-by-step logical inference using an LLM, avoiding the need for task-specific embedding optimization. Experiments on standard KG reasoning benchmarks demonstrate that ROG consistently outperforms strong embedding-based baselines in terms of mean reciprocal rank (MRR), with particularly notable gains on high-complexity query types. These results suggest that integrating structured KG retrieval with LLM-driven logical reasoning offers a robust and effective alternative for complex KG reasoning tasks.

cs.CL↗

arXiv2025

Kling-Omni Technical Report

We present Kling-Omni, a generalist generative framework designed to synthesize high-fidelity videos directly from multimodal visual language inputs. Adopting an end-to-end perspective, Kling-Omni bridges the functional separation among diverse video generation, editing, and intelligent reasoning tasks, integrating them into a holistic system. Unlike disjointed pipeline approaches, Kling-Omni supports a diverse range of user inputs, including text instructions, reference images, and video contexts, processing them into a unified multimodal representation to deliver cinematic-quality and highly-intelligent video content creation. To support these capabilities, we constructed a comprehensive data system that serves as the foundation for multimodal video creation. The framework is further empowered by efficient large-scale pre-training strategies and infrastructure optimizations for inference. Comprehensive evaluations reveal that Kling-Omni demonstrates exceptional capabilities in in-context generation, reasoning-based editing, and multimodal instruction following. Moving beyond a content creation tool, we believe Kling-Omni is a pivotal advancement toward multimodal world simulators capable of perceiving, reasoning, generating and interacting with the dynamic and complex worlds.

cs.CV↗

arXiv2025

KlingAvatar 2.0 Technical Report

Avatar video generation models have achieved remarkable progress in recent years. However, prior work exhibits limited efficiency in generating long-duration high-resolution videos, suffering from temporal drifting, quality degradation, and weak prompt following as video length increases. To address these challenges, we propose KlingAvatar 2.0, a spatio-temporal cascade framework that performs upscaling in both spatial resolution and temporal dimension. The framework first generates low-resolution blueprint video keyframes that capture global semantics and motion, and then refines them into high-resolution, temporally coherent sub-clips using a first-last frame strategy, while retaining smooth temporal transitions in long-form videos. To enhance cross-modal instruction fusion and alignment in extended videos, we introduce a Co-Reasoning Director composed of three modality-specific large language model (LLM) experts. These experts reason about modality priorities and infer underlying user intent, converting inputs into detailed storylines through multi-turn dialogue. A Negative Director further refines negative prompts to improve instruction alignment. Building on these components, we extend the framework to support ID-specific multi-character control. Extensive experiments demonstrate that our model effectively addresses the challenges of efficient, multimodally aligned long-form high-resolution video generation, delivering enhanced visual clarity, realistic lip-teeth rendering with accurate lip synchronization, strong identity preservation, and coherent multimodal instruction following.

cs.CV↗

arXiv2025

R^2-HGP: A Double-Regularized Gaussian Process for Heterogeneous Transfer Learning

Multi-output Gaussian process (MGP) models have attracted significant attention for their flexibility and uncertainty-quantification capabilities, and have been widely adopted in multi-source transfer learning scenarios due to their ability to capture inter-task correlations. However, they still face several challenges in transfer learning. First, the input spaces of the source and target domains are often heterogeneous, which makes direct knowledge transfer difficult. Second, potential prior knowledge and physical information are typically ignored during heterogeneous transfer, hampering the utilization of domain-specific insights and leading to unstable mappings. Third, inappropriate information sharing among target and sources can easily lead to negative transfer. Traditional models fail to address these issues in a unified way. To overcome these limitations, this paper proposes a Double-Regularized Heterogeneous Gaussian Process framework (R^2-HGP). Specifically, a trainable prior probability mapping model is first proposed to align the heterogeneous input domains. The resulting aligned inputs are treated as latent variables, upon which a multi-source transfer GP model is constructed and the entire structure is integrated into a novel conditional variational autoencoder (CVAE) based framework. Physical insights is further incorporated as a regularization term to ensure that the alignment results adhere to known physical knowledge. Next, within the multi-source transfer GP model, a sparsity penalty is imposed on the transfer coefficients, enabling the model to adaptively select the most informative source outputs and suppress negative transfer. Extensive simulations and real-world engineering case studies validate the effectiveness of our R^2-HGP, demonstrating consistent superiority over state-of-the-art benchmarks across diverse evaluation metrics.

cs.LG↗

arXiv2025

Spatial-spectral mapping for long-duration broadband terahertz pulse generation in on-chip waveguide arrays

Conventional approaches to terahertz (THz) pulse generation are restricted by the Fourier-transform limit, which hinders the creation of sources that combine long duration with broad bandwidth--a capability crucial for many spectroscopic and sensing applications. In this work, we overcome this challenge in the terahertz domain using an on-chip gradient waveguide array. The key is to spectrally disperse the pulse into spatially separated channels within a lithium niobate chip, effectively decoupling the design of temporal and spectral properties. We validate the source by distinguishing amino acid mixtures, demonstrating its tailored biosensing potential. This work establishes a novel mechanism for integrated THz generation, offering considerable promise for broadband spectroscopy and on-chip photonics.

physics.optics↗

arXiv2025

TCNN: Triple Convolutional Neural Network Models for Retrieval-based Question Answering System in E-commerce

Automatic question-answering (QA) systems have boomed during last few years, and commonly used techniques can be roughly categorized into Information Retrieval (IR)-based and generation-based. A key solution to the IR based models is to retrieve the most similar knowledge entries of a given query from a QA knowledge base, and then rerank those knowledge entries with semantic matching models. In this paper, we aim to improve an IR based e-commerce QA system-AliMe with proposed text matching models, including a basic Triple Convolutional Neural Network (TCNN) model and two Attention-based TCNN (ATCNN) models. Experimental results show their effect.

cs.LG↗

arXiv2025

AI Application in Anti-Money Laundering for Sustainable and Transparent Financial Systems

Money laundering and financial fraud remain major threats to global financial stability, costing trillions annually and challenging regulatory oversight. This paper reviews how artificial intelligence (AI) applications can modernize Anti-Money Laundering (AML) workflows by improving detection accuracy, lowering false-positive rates, and reducing the operational burden of manual investigations, thereby supporting more sustainable development. It further highlights future research directions including federated learning for privacy-preserving collaboration, fairness-aware and interpretable AI, reinforcement learning for adaptive defenses, and human-in-the-loop visualization systems to ensure that next-generation AML architectures remain transparent, accountable, and robust. In the final part, the paper proposes an AI-driven KYC application that integrates graph-based retrieval-augmented generation (RAG Graph) with generative models to enhance efficiency, transparency, and decision support in KYC processes related to money-laundering detection. Experimental results show that the RAG-Graph architecture delivers high faithfulness and strong answer relevancy across diverse evaluation settings, thereby enhancing the efficiency and transparency of KYC CDD/EDD workflows and contributing to more sustainable, resource-optimized compliance practices.

cs.AI↗

arXiv2025

Effect of Spin-Orbit Coupling on Anomalous Quantum Oscillations in InAs/GaSb Quantum Wells

We theoretically study the effect of spin-orbit coupling (SOC) on anomalous quantum oscillations in InAs/GaSb quantum wells. By comparing different cases, we show that SOC induces two opposing effects on anomalous quantum oscillations: it suppresses the oscillations in the clean case, while enhancing them in the disordered case. Using an effective model, we analyze in detail the origins of anomalous oscillations in both clean and disordered cases. Based on these origins, we explain why SOC suppresses or enhances the anomalous oscillations in different cases, thereby extending the understanding of the conventional theory. Moreover, in the disordered case, SOC can induce a phase shift of the anomalous oscillations. We further identify a parameter window where the anomalous oscillations are significantly enhanced in the presence of both disorder and SOC. These results provide a theoretical basis for understanding the role of SOC in anomalous quantum oscillations.

cond-mat.mes-hall↗

arXiv2025

GeoJSON Agents:A Multi-Agent LLM Architecture for Geospatial Analysis-Function Calling vs Code Generation

Large Language Models (LLMs) have demonstrated substantial progress in task automation and natural language understanding. However, without domain expertise in geographic information science (GIS), they continue to encounter limitations including reduced accuracy and unstable performance when processing complex tasks. To address these challenges, we propose GeoJSON Agents-a novel multi-agent LLM architecture specifically designed for geospatial analysis. This framework transforms natural language instructions into structured GeoJSON operations through two LLM enhancement techniques: Function Calling and Code Generation. The architecture integrates three core components: task parsing, agent collaboration, and result integration. The Planner agent systematically decomposes user-defined tasks into executable subtasks, while Worker agents perform spatial data processing and analysis either by invoking predefined function APIs or by generating and executing Python-based analytical code. The system produces reusable, standards-compliant GeoJSON outputs through iterative refinement. To evaluate both approaches, we constructed a benchmark comprising 70 tasks spanning basic, intermediate, and advanced complexity levels, conducting experiments with OpenAI's GPT-4o as the core model. Results indicate that the Code Generation-based agent achieved 97.14% accuracy, while the Function Calling-based agent attained 85.71%-both significantly outperforming the best-performing general-purpose model (48.57%). Comparative analysis reveals Code Generation offers superior flexibility for complex, open-ended tasks, whereas Function Calling provides enhanced execution stability for structured operations. This study represents the first systematic integration of GeoJSON data with a multi-agent LLM framework and provides empirical evidence comparing two mainstream enhancement methodologies in geospatial context.

cs.SE↗

arXiv2025

Stochastic fluctuations in an eco-evolutionary game dynamics with environmental feedbacks

Building upon the eco-evolutionary game dynamics framework established by Tilman et al., we investigate stochastic fluctuations in a two-strategy system incorporating environmental feedback mechanisms, where the payoff matrix exhibits population size dependence. We adopt a systematic approach which is the so-called $Ω$-expansion. When the stochastic factor is integrated, it is shown that the population size for each strategy fluctuates around the interior equilibrium of the macroscopic equations (corresponding to the deterministic model of the eco-evolutionary game) and its variance converges to a constant that is proportional to the environmental carrying capacity if the interior equilibrium is asymptotically stable. The simulation results demonstrate that the $Ω$ expansion provides a valid approximation, and the reliability of the aforementioned conclusions is verified. Therefore, analogous to Fudenberg and Harris' s stochastic replicator dynamics for infinite populations under external noise (\emph{J. Econ. Theory 57, 420-441}), the dynamic stability of the eco-evolutionary game can be extended to the stochastic regime when the environmental carrying capacity is sufficiently large.

q-bio.PE↗

arXiv2025

Quality analysis and evaluation prediction of RAG retrieval based on machine learning algorithms

With the rapid evolution of large language models, retrieval enhanced generation technology has been widely used due to its ability to integrate external knowledge to improve output accuracy. However, the performance of the system is highly dependent on the quality of the retrieval module. If the retrieval results have low relevance to user needs or contain noisy information, it will directly lead to distortion of the generated content. In response to the performance bottleneck of existing models in processing tabular features, this paper proposes an XGBoost machine learning regression model based on feature engineering and particle swarm optimization. Correlation analysis shows that answer_quality is positively correlated with doc_delevance by 0.66, indicating that document relevance has a significant positive effect on answer quality, and improving document relevance may enhance answer quality; The strong negative correlations between semantic similarity, redundancy, and diversity were -0.89 and -0.88, respectively, indicating a tradeoff between semantic similarity, redundancy, and diversity. In other words, as the former two increased, diversity significantly decreased. The experimental results comparing decision trees, AdaBoost, etc. show that the VMD PSO BiLSTM model is superior in all evaluation indicators, with significantly lower MSE, RMSE, MAE, and MAPE compared to the comparison model. The R2 value is higher, indicating that its prediction accuracy, stability, and data interpretation ability are more outstanding. This achievement provides an effective path for optimizing the retrieval quality and improving the generation effect of RAG system, and has important value in promoting the implementation and application of related technologies.

cs.LG↗

arXiv2025

Equivariant embeddings of Riemann surfaces in Euclidean spaces with minimal dimensions

Let $Σ_g$ be a closed Riemann surface of genus $g$. Let $G$ be a finite subgroup of the automorphism group of $Σ_g$. It is well known that there exists a smooth $G$-equivariant embedding from $Σ_g$ to some Euclidean space $\mathbb{R}^n$. Let $d_g(G)$ be the minimal possible $n$ for $(Σ_g,G)$. We compute the value of $d_g(G)$ in certain cases. Especially, we show that: for the automorphism group of the closed Riemann surface which comes from the principal congruence subgroup of level $p$, where $p\geq 7$ is prime, $d_g(G)=p+1$. As a corollary, the minimal $n$ for the Hurwitz action on the Klein quartic is equal to $8$. Three kinds of methods are used in the computation, which are related to the representations of groups, the equivariant triangulations, and the orbifold theory, respectively. The methods are also used to provide two kinds of upper bounds: $d_g(G)\leq |G|$ if $|G|\geq 5$; and $d_g(G)\leq 12(g-1)$ if $g\geq 2$.

math.GT↗

arXiv2025

Gaussian Splatting-based Low-Rank Tensor Representation for Multi-Dimensional Image Recovery

Tensor singular value decomposition (t-SVD) is a promising tool for multi-dimensional image representation, which decomposes a multi-dimensional image into a latent tensor and an accompanying transform matrix. However, two critical limitations of t-SVD methods persist: (1) the approximation of the latent tensor (e.g., tensor factorizations) is coarse and fails to accurately capture spatial local high-frequency information; (2) The transform matrix is composed of fixed basis atoms (e.g., complex exponential atoms in DFT and cosine atoms in DCT) and cannot precisely capture local high-frequency information along the mode-3 fibers. To address these two limitations, we propose a Gaussian Splatting-based Low-rank tensor Representation (GSLR) framework, which compactly and continuously represents multi-dimensional images. Specifically, we leverage tailored 2D Gaussian splatting and 1D Gaussian splatting to generate the latent tensor and transform matrix, respectively. The 2D and 1D Gaussian splatting are indispensable and complementary under this representation framework, which enjoys a powerful representation capability, especially for local high-frequency information. To evaluate the representation ability of the proposed GSLR, we develop an unsupervised GSLR-based multi-dimensional image recovery model. Extensive experiments on multi-dimensional image recovery demonstrate that GSLR consistently outperforms state-of-the-art methods, particularly in capturing local high-frequency information.

cs.CV↗