arXiv subjects

Yan Wang

Publications and source records attributed to Yan Wang.

At least 325 records · Page 18Linked to original sources

arXiv2025

Reducing Systematic Bias in Machine Learning Applications to Signal Extraction in High-Energy Nuclear Physics

Machine learning techniques are increasingly being applied in high-energy nuclear physics data analysis thanks to their outstanding performance. One key challenge in such applications is the construction of training samples that can accurately represent real data. Training samples are typically generated through detector simulations, but discrepancies between simulated and real data can lead to degradation in machine learning performance and systematic biases in the results. This paper introduces two methods: i) cumulative distribution function mapping and ii) shift-and-scale, to align simulated signals with real data, which can aid in eliminating the aforementioned issues. We use the J/$ψ$ yield measurement in 200 GeV Ru+Ru and Zr+Zr collisions with the STAR experiment as an example to demonstrate the application and effectiveness of the proposed methods.

nucl-ex↗

arXiv2025

DualFete: Revisiting Teacher-Student Interactions from a Feedback Perspective for Semi-supervised Medical Image Segmentation

The teacher-student paradigm has emerged as a canonical framework in semi-supervised learning. When applied to medical image segmentation, the paradigm faces challenges due to inherent image ambiguities, making it particularly vulnerable to erroneous supervision. Crucially, the student's iterative reconfirmation of these errors leads to self-reinforcing bias. While some studies attempt to mitigate this bias, they often rely on external modifications to the conventional teacher-student framework, overlooking its intrinsic potential for error correction. In response, this work introduces a feedback mechanism into the teacher-student framework to counteract error reconfirmations. Here, the student provides feedback on the changes induced by the teacher's pseudo-labels, enabling the teacher to refine these labels accordingly. We specify that this interaction hinges on two key components: the feedback attributor, which designates pseudo-labels triggering the student's update, and the feedback receiver, which determines where to apply this feedback. Building on this, a dual-teacher feedback model is further proposed, which allows more dynamics in the feedback loop and fosters more gains by resolving disagreements through cross-teacher supervision while avoiding consistent errors. Comprehensive evaluations on three medical image benchmarks demonstrate the method's effectiveness in addressing error propagation in semi-supervised medical image segmentation.

cs.CV↗

arXiv2025

WebVIA: A Web-based Vision-Language Agentic Framework for Interactive and Verifiable UI-to-Code Generation

User interface (UI) development requires translating design mockups into functional code, a process that remains repetitive and labor-intensive. While recent Vision-Language Models (VLMs) automate UI-to-Code generation, they generate only static HTML/CSS/JavaScript layouts lacking interactivity. To address this, we propose WebVIA, the first agentic framework for interactive UI-to-Code generation and validation. The framework comprises three components: 1) an exploration agent to capture multi-state UI screenshots; 2) a UI2Code model that generates executable interactive code; 3) a validation module that verifies the interactivity. Experiments demonstrate that WebVIA-Agent achieves more stable and accurate UI exploration than general-purpose agents (e.g., Gemini-2.5-Pro). In addition, our fine-tuned WebVIA-UI2Code models exhibit substantial improvements in generating executable and interactive HTML/CSS/JavaScript code, outperforming their base counterparts across both interactive and static UI2Code benchmarks. Our code and models are available at \href{https://zheny2751-dotcom.github.io/webvia.github.io/}{\texttt{https://webvia.github.io}}.

cs.SE↗

arXiv2025

V-Shuffle: Zero-Shot Style Transfer via Value Shuffle

Attention injection-based style transfer has achieved remarkable progress in recent years. However, existing methods often suffer from content leakage, where the undesired semantic content of the style image mistakenly appears in the stylized output. In this paper, we propose V-Shuffle, a zero-shot style transfer method that leverages multiple style images from the same style domain to effectively navigate the trade-off between content preservation and style fidelity. V-Shuffle implicitly disrupts the semantic content of the style images by shuffling the value features within the self-attention layers of the diffusion model, thereby preserving low-level style representations. We further introduce a Hybrid Style Regularization that complements these low-level representations with high-level style textures to enhance style fidelity. Empirical results demonstrate that V-Shuffle achieves excellent performance when utilizing multiple style images. Moreover, when applied to a single style image, V-Shuffle outperforms previous state-of-the-art methods.

cs.CV↗

arXiv2025

Crossover from quantum correlation to hot-carrier transport in scattering-tolerant 2D transistors

Quantum correlation and hot-carrier transport represent two fundamentally distinct regimes of electronic conduction, rarely accessible within the same device. Here, we report a state-of-the-art monolayer transition metal dichalcogenides transistor architecture on a ferroelectric substrate that enables this crossover by leveraging the strong dielectric screening and in-plane gate control. At cryogenic temperatures, the devices exhibit reproducible quasi-periodic current fluctuations, consistent with an emergent potential landscape driven by electron-electron interactions at low carrier densities. As the temperature increases, this correlated potential profile thermally dissolves and transport is dominated by the lateral gate-field that drives the carriers with high kinetic energy. These hot-carriers can efficiently surmount the scattering events, exhibiting a record-high room-temperature electron mobility of ~4,800 cm^2/Vs and a maximum on-current ~0.5 mA/μm, surpassing traditional FETs in key performance metrics. These findings establish a unified approach for probing intermediate mesoscopic orders, while advancing the transistor performance limits in scalable 2D transistors.

cond-mat.mes-hall↗

arXiv2025

Co-Designing Spectral Transformation Oracles with Hybrid Oscillator-Qubit Quantum Processors: From Algorithms to Compilation

We co-design a family of quantum eigenvalue transformation oracles that can be efficiently implemented on hybrid discrete/continuous-variable (qubit/qumode) hardware. To illustrate the oracle's representation-theoretic power and near-term experimental accessibility, we encode a Gaussian imaginary time evolution spectral filter. As a result, we define a continuous linear combination of unitaries block-encoding. Due to the ancillary qumode's infinite-dimensional nature, continuous variable qumodes constitute a powerful compilation tool for encoding continuous spectral functions without discretization errors while minimizing resource requirements. We then focus on the ubiquitous task of preparing eigenstates in quantum spin models. For completeness, we provide an end-to-end compilation which expresses high-level oracles in terms of an experimentally realizable instruction set architecture in both 1D and 2D. Finally, we examine the leading-order effects of physical errors and highlight open research directions. Our algorithms scale linearly with the spatial extent of the target system and are applicable to both near-term and large-scale quantum processors.

quant-ph↗

arXiv2025

Geographic variability in reanalysis wind speed biases: A high-resolution bias correction approach for UK wind energy

Reanalysis datasets have become indispensable tools for wind resource assessment and wind power simulation, offering long-term and spatially continuous wind fields across large regions. However, they inherently contain systematic wind speed biases arising from various factors, including simplified physical parameterizations, observational uncertainties, and limited spatial resolution. Among these, low spatial resolution poses a particular challenge for capturing local variability accurately. Whereas prevailing industry practice generally relies on either no bias correction or coarse, nationally uniform adjustments, we extend and thoroughly analyse a recently proposed spatially resolved, cluster-based bias correction framework. This approach is designed to better account for local heterogeneity and is applied to 319 wind farms across the United Kingdom to evaluate its effectiveness. Results show that this method reduced monthly wind power simulation errors by more than 32% compared to the uncorrected ERA5 reanalysis dataset. The method is further applied to the MERRA-2 dataset for comparative evaluation, demonstrating its effectiveness and robustness for different reanalysis products. In contrast to prior studies, which rarely quantify the influence of topography on reanalysis biases, this research presents a detailed spatial mapping of bias correction factors across the UK. The analysis reveals that for wind energy applications, ERA5 wind speed errors exhibit strong spatial variability, with the most significant underestimations in the Scottish Highlands and mountainous areas of Wales. These findings highlight the importance of explicitly accounting for geographic variability when correcting reanalysis wind speeds, and provide new insights into region-specific bias patterns relevant for high-resolution wind energy modelling.

physics.ao-ph↗

arXiv2025

The End of Manual Decoding: Towards Truly End-to-End Language Models

The "end-to-end" label for LLMs is a misnomer. In practice, they depend on a non-differentiable decoding process that requires laborious, hand-tuning of hyperparameters like temperature and top-p. This paper introduces AutoDeco, a novel architecture that enables truly "end-to-end" generation by learning to control its own decoding strategy. We augment the standard transformer with lightweight heads that, at each step, dynamically predict context-specific temperature and top-p values alongside the next-token logits. This approach transforms decoding into a parametric, token-level process, allowing the model to self-regulate its sampling strategy within a single forward pass. Through extensive experiments on eight benchmarks, we demonstrate that AutoDeco not only significantly outperforms default decoding strategies but also achieves performance comparable to an oracle-tuned baseline derived from "hacking the test set"-a practical upper bound for any static method. Crucially, we uncover an emergent capability for instruction-based decoding control: the model learns to interpret natural language commands (e.g., "generate with low randomness") and adjusts its predicted temperature and top-p on a token-by-token basis, opening a new paradigm for steerable and interactive LLM decoding.

cs.CL↗

arXiv2025

When Agents Trade: Live Multi-Market Trading Benchmark for LLM Agents

Although Large Language Model (LLM)-based agents are increasingly used in financial trading, it remains unclear whether they can reason and adapt in live markets, as most studies test models instead of agents, cover limited periods and assets, and rely on unverified data. To address these gaps, we introduce Agent Market Arena (AMA), the first lifelong, real-time benchmark for evaluating LLM-based trading agents across multiple markets. AMA integrates verified trading data, expert-checked news, and diverse agent architectures within a unified trading framework, enabling fair and continuous comparison under real conditions. It implements four agents, including InvestorAgent as a single-agent baseline, TradeAgent and HedgeFundAgent with different risk styles, and DeepFundAgent with memory-based reasoning, and evaluates them across GPT-4o, GPT-4.1, Claude-3.5-haiku, Claude-sonnet-4, and Gemini-2.0-flash. Live experiments on both cryptocurrency and stock markets demonstrate that agent frameworks display markedly distinct behavioral patterns, spanning from aggressive risk-taking to conservative decision-making, whereas model backbones contribute less to outcome variation. AMA thus establishes a foundation for rigorous, reproducible, and continuously evolving evaluation of financial reasoning and trading intelligence in LLM-based agents.

cs.CL↗

arXiv2025

OneRec-V2 Technical Report

Recent breakthroughs in generative AI have transformed recommender systems through end-to-end generation. OneRec reformulates recommendation as an autoregressive generation task, achieving high Model FLOPs Utilization. While OneRec-V1 has shown significant empirical success in real-world deployment, two critical challenges hinder its scalability and performance: (1) inefficient computational allocation where 97.66% of resources are consumed by sequence encoding rather than generation, and (2) limitations in reinforcement learning relying solely on reward models. To address these challenges, we propose OneRec-V2, featuring: (1) Lazy Decoder-Only Architecture: Eliminates encoder bottlenecks, reducing total computation by 94% and training resources by 90%, enabling successful scaling to 8B parameters. (2) Preference Alignment with Real-World User Interactions: Incorporates Duration-Aware Reward Shaping and Adaptive Ratio Clipping to better align with user preferences using real-world feedback. Extensive A/B tests on Kuaishou demonstrate OneRec-V2's effectiveness, improving App Stay Time by 0.467%/0.741% while balancing multi-objective recommendations. This work advances generative recommendation scalability and alignment with real-world feedback, representing a step forward in the development of end-to-end recommender systems.

cs.IR↗

arXiv2025

OMPQ: Orthogonal Mixed Precision Quantization

To bridge the ever increasing gap between deep neural networks' complexity and hardware capability, network quantization has attracted more and more research attention. The latest trend of mixed precision quantization takes advantage of hardware's multiple bit-width arithmetic operations to unleash the full potential of network quantization. However, this also results in a difficult integer programming formulation, and forces most existing approaches to use an extremely time-consuming search process even with various relaxations. Instead of solving a problem of the original integer programming, we propose to optimize a proxy metric, the concept of network orthogonality, which is highly correlated with the loss of the integer programming but also easy to optimize with linear programming. This approach reduces the search time and required data amount by orders of magnitude, with little compromise on quantization accuracy. Specifically, we achieve 72.08% Top-1 accuracy on ResNet-18 with 6.7Mb, which does not require any searching iterations. Given the high efficiency and low data dependency of our algorithm, we used it for the post-training quantization, which achieve 71.27% Top-1 accuracy on MobileNetV2 with only 1.5Mb. Our code is available at https://github.com/MAC-AutoML/OMPQ.

cs.LG↗

arXiv2025

Addressing Mark Imbalance in Integration-free Neural Marked Temporal Point Processes

Marked Temporal Point Process (MTPP) has been well studied to model the event distribution in marked event streams, which can be used to predict the mark and arrival time of the next event. However, existing studies overlook that the distribution of event marks is highly imbalanced in many real-world applications, with some marks being frequent but others rare. The imbalance poses a significant challenge to the performance of the next event prediction, especially for events of rare marks. To address this issue, we propose a thresholding method, which learns thresholds to tune the mark probability normalized by the mark's prior probability to optimize mark prediction, rather than predicting the mark directly based on the mark probability as in existing studies. In conjunction with this method, we predict the mark first and then the time. In particular, we develop a novel neural MTPP model to support effective time sampling and estimation of mark probability without computationally expensive numerical improper integration. Extensive experiments on real-world datasets demonstrate the superior performance of our solution against various baselines for the next event mark and time prediction. The code is available at https://github.com/undes1red/IFNMTPP.

cs.LG↗

arXiv2025

Simulation of Repair on Dynamic Patient-Specific Left Atrioventricular Valve Models

Purpose: To develop and evaluate a dynamic, image-derived patient-specific physical simulation platform for the assessment of left atrioventricular valve (LAVV) repair strategies in pediatric patients with repaired atrioventricular canal defects. Methods: 3D transesophageal echocardiographic images of two patients with regurgitant LAVVs were identified from an institutional database. Custom code in SlicerHeart was used to segment leaflets, define the annulus, and generate patient-specific valve molds. Silicone valve models were fabricated and tested in a pulse duplicator under simulated physiological conditions. Five unrepaired valves were analyzed for manufacturing consistency, and multiple surgical repair techniques were compared for two patient-specific models. Results: Manufacturing variability was low in annular metrics (CV for annular circumference: 2.1; commissural distance: 4.1; annulus height: 14.7) but higher in leaflet closure metrics (billow height: 11.1%; billow volume: 18.9%; tenting height: 45.9%; tenting volume: 73.5%). In Patient 1, cleft closure and an Alfieri stitch both eliminated the regurgitant orifice area, but the Alfieri stitch resulted in elevated mean pressure gradient (17 mmHg vs. 4-9 mmHg for other repairs) and deteriorated with repeated loading. In Patient 2, no repair eliminated regurgitation entirely; however, combining an 11 mm patch augmentation with commissuroplasty reduced regurgitant area to 0.147 cm2, the smallest observed among tested strategies. Conclusion: This study demonstrates the feasibility of a dynamic physical simulation platform for preclinical LAVV repair evaluation. This proof-of-concept work highlights the platform potential for refining repair strategies before clinical application, which may be particularly relevant in small and heterogeneous populations with congenital heart disease.

physics.med-ph↗

arXiv2025

Non-Parametric Estimation Techniques of Factor Copula Model using Proxies

Parametric factor copula models typically work well in modeling multivariate dependencies due to their flexibility and ability to capture complex dependency structures. However, accurately estimating the linking copulas within these models remains challenging, especially when working with high-dimensional data. This paper proposes a novel approach for estimating linking copulas based on a non-parametric kernel estimator. Unlike conventional parametric methods, our approach utilizes the flexibility of kernel density estimation to capture the underlying dependencies more accurately, particularly in scenarios where the underlying copula structure is complex or unknown. We show that the proposed estimator is consistent under mild conditions and demonstrate its effectiveness through extensive simulation studies. Our findings suggest that the proposed approach offers a promising avenue for modeling multivariate dependencies, particularly in applications requiring robust and efficient estimation of copula-based models.

stat.ME↗

arXiv2025

Efficiency-Enhanced Open Earbud Earphone Antenna Using Dual-Feed Technique

The stringent spatial constraints and the demand for high antenna efficiency in modern wireless earphones present significant design challenges. To address these issues, this paper presents and thoroughly investigates a novel earphone antenna design specifically tailored for open earbud wireless earphones. In contrast to traditional earphone antennas that rely on a conventional single-feed configuration, the proposed design introduces a dual-feed excitation technique incorporating a controlled phase difference between the two feeds. This innovative feeding strategy effectively enlarges the equivalent radiating aperture, thereby enhancing the overall radiation efficiency of the antenna system. Experimental and simulation results demonstrate that the dual-feed approach yields an efficiency improvement exceeding 1 dB when compared with standard single-feed designs. Furthermore, the fabricated prototype achieves a -6 dB impedance bandwidth that fully encompasses the 2.4 GHz ISM band, ensuring stable wireless communication performance. The measured total efficiencies reach -8.5 dB in free space and -9.5 dB under on-head conditions. These results confirm that the proposed antenna successfully achieves high efficiency and reliable performance within the extremely limited volume of an earbud device, demonstrating strong potential for integration into next-generation compact wireless earphones.

eess.SP↗

arXiv2025

F-Adapter: Frequency-Adaptive Parameter-Efficient Fine-Tuning in Scientific Machine Learning

Parameter-efficient fine-tuning (PEFT) of powerful pre-trained models for complex downstream tasks has proven effective in vision and language processing, yet this paradigm remains unexplored in scientific machine learning, where the objective is to model complex physical systems. We conduct the first systematic study of PEFT for pre-trained Large Operator Models (LOMs) obtained by scaling variants of Fourier Neural Operator. First, we observe that the widely used Low-Rank Adaptation (LoRA) yields markedly poorer performance on LOMs than Adapter tuning. Then, we further theoretically establish that stacked LoRA incurs a depth-amplified lower bound on approximation error within Fourier layers, whereas adapters retain universal approximation capacity and, by concentrating parameters on energy-dominant low-frequency modes, attain exponentially decaying error with bottleneck width in the Fourier domain. Motivated by the robust empirical gains of adapters and by our theoretical characterization of PDE solutions as spectrally sparse, we introduce Frequency-Adaptive Adapter (F-Adapter). F-Adapter allocates adapter capacity based on spectral complexity, assigning higher-dimension modules to low-frequency components and lower-dimension modules to high-frequency components. Our F-Adapters establish state-of-the-art (SOTA) results on multiple challenging 3D Navier-Stokes benchmarks, markedly enhancing both generalization and spectral fidelity over LoRA and other PEFT techniques commonly used in LLMs. To the best of our knowledge, this work is the first to explore PEFT for scientific machine-learning and establishes F-Adapter as an effective paradigm for this domain.

cs.LG↗

arXiv2025

Video-STR: Reinforcing MLLMs in Video Spatio-Temporal Reasoning with Relation Graph

Recent progress in Multimodal Large Language Models (MLLMs) has demonstrated strong semantic understanding capabilities, but struggles to perform precise spatio-temporal understanding. Existing spatio-temporal methods primarily focus on the video itself, while overlooking the physical information within the video, such as multi-object layouts and motion. Such limitations restrict the use of MLLMs in downstream applications that demand high precision, including embodied intelligence and VR. To address this issue, we present Video-STR, a novel graph-based reinforcement method for precise Video Spatio-Temporal Reasoning. Building upon the capacity of Reinforcement Learning with Verifiable Reward (RLVR) to improve model abilities, we introduce a reasoning mechanism using graph-based Group Relative Policy Optimization (GRPO) method to guide the model in inferring the underlying spatio-temporal topology of scenarios during the thinking process. To resolve the lack of spatio-temporal training data, we construct the STV-205k dataset with 205k question-answering pairs, covering dynamic multi-object scenes in both indoor and outdoor environments, to support the model training. Experiments show that Video-STR achieves state-of-the-art results on various benchmarks, outperforming the base model by 13% on STI-Bench, and demonstrating the effectiveness of our approach and dataset. Code, model, and data will be released.

cs.AI↗

arXiv2025

MultiFinBen: Benchmarking Large Language Models for Multilingual and Multimodal Financial Application

Real-world financial analysis involves information across multiple languages and modalities, from reports and news to scanned filings and meeting recordings. Yet most existing evaluations of LLMs in finance remain text-only, monolingual, and largely saturated by current models. To bridge these gaps, we present MultiFinBen, the first expert-annotated multilingual (five languages) and multimodal (text, vision, audio) benchmark for evaluating LLMs in realistic financial contexts. MultiFinBen introduces two new task families: multilingual financial reasoning, which tests cross-lingual evidence integration from filings and news, and financial OCR, which extracts structured text from scanned documents containing tables and charts. Rather than aggregating all available datasets, we apply a structured, difficulty-aware selection based on advanced model performance, ensuring balanced challenge and removing redundant tasks. Evaluating 21 leading LLMs shows that even frontier multimodal models like GPT-4o achieve only 46.01% overall, stronger on vision and audio but dropping sharply in multilingual settings. These findings expose persistent limitations in multilingual, multimodal, and expert-level financial reasoning. All datasets, evaluation scripts, and leaderboards are publicly released.

cs.CL↗