arXiv subjects

Bin Yan

Publications and source records attributed to Bin Yan.

At least 37 records · Page 2Linked to original sources

arXiv2025

Probing top-quark electroweak couplings indirectly at the Electron-Ion Collider

Top quark electroweak interactions serve as a sensitive probe for beyond-Standard-Model physics, potentially exhibiting significant deviations from Standard Model predictions and offering crucial tests of fundamental symmetries in ultraviolet physics. This Letter investigates their measurement through deep inelastic scattering (DIS) processes involving top quark loops at the Electron-Ion Collider (EIC), within the Standard Model Effective Field Theory framework. We demonstrate that left-handed electron beam polarization can significantly enhance DIS cross sections for these interactions compared to right-handed scenario, highlighting the pivotal role of high polarization at the EIC in constraining these couplings. Compared to direct $pp \to t\bar{t}Z$ measurements at the Large Hadron Collider (LHC), DIS measurements could significantly improve these coupling constraints and effectively resolve parameter space degeneracies present in LHC data. Moreover, the expected precision from the EIC's high-luminosity phase matches that of LEP electroweak precision measurements, underscoring the EIC's significant potential to probe top quark properties.

hep-ph↗

arXiv2025

InfinityStar: Unified Spacetime AutoRegressive Modeling for Visual Generation

We introduce InfinityStar, a unified spacetime autoregressive framework for high-resolution image and dynamic video synthesis. Building on the recent success of autoregressive modeling in both vision and language, our purely discrete approach jointly captures spatial and temporal dependencies within a single architecture. This unified design naturally supports a variety of generation tasks such as text-to-image, text-to-video, image-to-video, and long interactive video synthesis via straightforward temporal autoregression. Extensive experiments demonstrate that InfinityStar scores 83.74 on VBench, outperforming all autoregressive models by large margins, even surpassing some diffusion competitors like HunyuanVideo. Without extra optimizations, our model generates a 5s, 720p video approximately 10x faster than leading diffusion-based methods. To our knowledge, InfinityStar is the first discrete autoregressive video generator capable of producing industrial level 720p videos. We release all code and models to foster further research in efficient, high-quality video generation.

cs.CV↗

arXiv2025

Multi-Text Guided Few-Shot Semantic Segmentation

Recent CLIP-based few-shot semantic segmentation methods introduce class-level textual priors to assist segmentation by typically using a single prompt (e.g., a photo of class). However, these approaches often result in incomplete activation of target regions, as a single textual description cannot fully capture the semantic diversity of complex categories. Moreover, they lack explicit cross-modal interaction and are vulnerable to noisy support features, further degrading visual prior quality. To address these issues, we propose the Multi-Text Guided Few-Shot Semantic Segmentation Network (MTGNet), a dual-branch framework that enhances segmentation performance by fusing diverse textual prompts to refine textual priors and guide the cross-modal optimization of visual priors. Specifically, we design a Multi-Textual Prior Refinement (MTPR) module that suppresses interference and aggregates complementary semantic cues to enhance foreground activation and expand semantic coverage for structurally complex objects. We introduce a Text Anchor Feature Fusion (TAFF) module, which leverages multi-text embeddings as semantic anchors to facilitate the transfer of discriminative local prototypes from support images to query images, thereby improving semantic consistency and alleviating intra-class variations. Furthermore, a Foreground Confidence-Weighted Attention (FCWA) module is presented to enhance visual prior robustness by leveraging internal self-similarity within support foreground features. It adaptively down-weights inconsistent regions and effectively suppresses interference in the query segmentation process. Extensive experiments on standard FSS benchmarks validate the effectiveness of MTGNet. In the 1-shot setting, it achieves 76.8% mIoU on PASCAL-5i and 57.4% on COCO-20i, with notable improvements in folds exhibiting high intra-class variations.

cs.CV↗

arXiv2025

MFI-ResNet: Efficient ResNet Architecture Optimization via MeanFlow Compression and Selective Incubation

ResNet has achieved tremendous success in computer vision through its residual connection mechanism. ResNet can be viewed as a discretized form of ordinary differential equations (ODEs). From this perspective, the multiple residual blocks within a single ResNet stage essentially perform multi-step discrete iterations of the feature transformation for that stage. The recently proposed flow matching model, MeanFlow, enables one-step generative modeling by learning the mean velocity field to transform distributions. Inspired by this, we propose MeanFlow-Incubated ResNet (MFI-ResNet), which employs a compression-expansion strategy to jointly improve parameter efficiency and discriminative performance. In the compression phase, we simplify the multi-layer structure within each ResNet stage to one or two MeanFlow modules to construct a lightweight meta model. In the expansion phase, we apply a selective incubation strategy to the first three stages, expanding them to match the residual block configuration of the baseline ResNet model, while keeping the last stage in MeanFlow form, and fine-tune the incubated model. Experimental results show that on CIFAR-10 and CIFAR-100 datasets, MFI-ResNet achieves remarkable parameter efficiency, reducing parameters by 46.28% and 45.59% compared to ResNet-50, while still improving accuracy by 0.23% and 0.17%, respectively. This demonstrates that generative flow-fields can effectively characterize the feature transformation process in ResNet, providing a new perspective for understanding the relationship between generative modeling and discriminative learning.

cs.CV↗

arXiv2025

Transverse spin effects and light-quark dipole moments at lepton colliders

We propose to probe light-quark dipole interactions at lepton colliders using the azimuthal asymmetry of a collinear dihadron pair $(h_1h_2)$ produced in association with another hadron $h'$. This asymmetry, arising from quantum interference in the quark spin space, is exclusively sensitive to dipole interactions at the leading power of the new physics scale and simultaneously probes both the real and imaginary components of the dipole couplings. By combining all possible channels of $h'$, this method allows for disentangling the up and down quark dipole moments and has the potential to significantly strengthen current constraints by one to two orders of magnitude.

hep-ph↗

arXiv2025

Injecting Domain-Specific Knowledge into Large Language Models: A Comprehensive Survey

Large Language Models (LLMs) have demonstrated remarkable success in various tasks such as natural language understanding, text summarization, and machine translation. However, their general-purpose nature often limits their effectiveness in domain-specific applications that require specialized knowledge, such as healthcare, chemistry, or legal analysis. To address this, researchers have explored diverse methods to enhance LLMs by integrating domain-specific knowledge. In this survey, we provide a comprehensive overview of these methods, which we categorize into four key approaches: dynamic knowledge injection, static knowledge embedding, modular adapters, and prompt optimization. Each approach offers unique mechanisms to equip LLMs with domain expertise, balancing trade-offs between flexibility, scalability, and efficiency. We discuss how these methods enable LLMs to tackle specialized tasks, compare their advantages and disadvantages, evaluate domain-specific LLMs against general LLMs, and highlight the challenges and opportunities in this emerging field. For those interested in delving deeper into this area, we also summarize the commonly used datasets and benchmarks. To keep researchers updated on the latest studies, we maintain an open-source at: https://github.com/abilliyb/Knowledge_Injection_Survey_Papers, dedicated to documenting research in the field of specialized LLM.

cs.CL↗

arXiv2025

QuarkMed Medical Foundation Model Technical Report

Recent advancements in large language models have significantly accelerated their adoption in healthcare applications, including AI-powered medical consultations, diagnostic report assistance, and medical search tools. However, medical tasks often demand highly specialized knowledge, professional accuracy, and customization capabilities, necessitating a robust and reliable foundation model. QuarkMed addresses these needs by leveraging curated medical data processing, medical-content Retrieval-Augmented Generation (RAG), and a large-scale, verifiable reinforcement learning pipeline to develop a high-performance medical foundation model. The model achieved 70% accuracy on the Chinese Medical Licensing Examination, demonstrating strong generalization across diverse medical benchmarks. QuarkMed offers a powerful yet versatile personal medical AI solution, already serving over millions of users at ai.quark.cn.

cs.AI↗

arXiv2025

Extracting randomness from magic quantum states

Magic quantum states (non-stabilizer states) play a pivotal role in fault-tolerant quantum computation. Simultaneously, random resources have emerged as a key element in various randomized techniques within contemporary quantum science. In this study, we establish a direct connection between these two notions. More specifically, our research demonstrates that when a subsystem of a quantum state is measured, the resultant unmeasured part of the system can exhibit a high degree of randomness that can be enhanced by the inherent correlations of the underlying magic quantum state. Our findings suggest an approach to quantifying correlations within magic quantum states beyond the conventional paradigm of entanglement, and introduce an efficient approach for leveraging such correlations to generate random quantum resources.

quant-ph↗

arXiv2025

Bell Inequality Violation of Light Quarks in Back-to-Back Dihadron Pair Production at Lepton Colliders

Spin correlations between particles produced at colliders provide valuable insights for quantum information studies. While traditional studies of quantum information at colliders are typically limited to massive particles with perturbative decay, we propose an innovative method to explore the Bell inequality in massless quark pair systems by analyzing the azimuthal correlations in back-to-back $π^+π^-$ dihadron pair production at lepton colliders. Revisiting the Belle data, we have shown the potential to detect Bell inequality violation of light quarks by introducing an additional angular cut, achieving a significance of 2.5 $σ$ even in the worst-case scenario of 100% correlated systematic uncertainties in each bins. The significance substantially exceeds $5σ$ when considering uncorrelated systematic uncertainties. Our approach opens avenues for exploring spin quantum information in the non-perturbative aspect and leverages existing data for quantum information research.

hep-ph↗

arXiv2025

Origin-Destination Travel Demand Estimation: An Approach That Scales Worldwide, and Its Application to Five Metropolitan Highway Networks

Estimating Origin-Destination (OD) travel demand is vital for effective urban planning and traffic management. Developing universally applicable OD estimation methodologies is significantly challenged by the pervasive scarcity of high-fidelity traffic data and the difficulty in obtaining city-specific prior OD estimates (or seed ODs), which are often prerequisite for traditional approaches. Our proposed method directly estimates OD travel demand by systematically leveraging aggregated, anonymized statistics from Google Maps Traffic Trends, obviating the need for conventional census or city-provided OD data. The OD demand is estimated by formulating a single-level, one-dimensional, continuous nonlinear optimization problem with nonlinear equality and bound constraints to replicate highway path travel times. The method achieves efficiency and scalability by employing a differentiable analytical macroscopic network model. This model by design is computationally lightweight, distinguished by its parsimonious parameterization that requires minimal calibration effort and its capacity for instantaneous evaluation. These attributes ensure the method's broad applicability and practical utility across diverse cities globally. Using segment sensor counts from Los Angeles and San Diego highway networks, we validate our proposed approach, demonstrating a two-thirds to three-quarters improvement in the fit to segment count data over a baseline. Beyond validation, we establish the method's scalability and robust performance in replicating path travel times across diverse highway networks, including Seattle, Orlando, Denver, Philadelphia, and Boston. In these expanded evaluations, our method not only aligns with simulation-based benchmarks but also achieves an average 13% improvement in it's ability to fit travel time data compared to the baseline during afternoon peak hours.

cs.ET↗

arXiv2025

Testing Lepton Flavor Universality at the Electron-Ion Collider

Measurements of $b\to cτ^-\barν_τ$ transitions at colliders are highly motivated for testing lepton flavor universality (LFU), a cornerstone hypothesis of the Standard Model (SM). Potential observations of LFU violation could provide significant evidence for physics beyond the SM (BSM). The substantial production of $b$-hadrons at the Electron-Ion Collider (EIC) would highlight its potential to support LFU testing and complement studies conducted at other experimental facilities. In this paper, we study the production of $b$-hadrons in deep inelastic scattering processes at the EIC with $\sqrt{s}=100\,\rm GeV$. We estimate the $b$-hadron yields at the EIC to reach $\mathcal O(10^9)$ with an integrated luminosity of up to $10^3\,{\rm fb}^{-1}$. Furthermore, we perform a systematic study on various $b$-hadron decays, exploring the sensitivities of LFU-violating observables, including $R_{J/ψ}$, $R_{D_s^{(\ast)}}$ and $R_{Λ_c}$. Detailed strategies for track-based event reconstruction are investigated for these observables. We also include a discussion of the annihilation process $B_c^+ \to τ^+ν_τ$. Finally, we provide a theoretical interpretation of the projected sensitivities within the framework of low energy effective field theory (LEFT). Our analysis indicates that the EIC has the potential to constrain the relevant Wilson coefficients at $\mathcal O(0.1)$, offering complementary insights to other measurements.

hep-ph↗

arXiv2025

Infinity: Scaling Bitwise AutoRegressive Modeling for High-Resolution Image Synthesis

We present Infinity, a Bitwise Visual AutoRegressive Modeling capable of generating high-resolution, photorealistic images following language instruction. Infinity redefines visual autoregressive model under a bitwise token prediction framework with an infinite-vocabulary tokenizer & classifier and bitwise self-correction mechanism, remarkably improving the generation capacity and details. By theoretically scaling the tokenizer vocabulary size to infinity and concurrently scaling the transformer size, our method significantly unleashes powerful scaling capabilities compared to vanilla VAR. Infinity sets a new record for autoregressive text-to-image models, outperforming top-tier diffusion models like SD3-Medium and SDXL. Notably, Infinity surpasses SD3-Medium by improving the GenEval benchmark score from 0.62 to 0.73 and the ImageReward benchmark score from 0.87 to 0.96, achieving a win rate of 66%. Without extra optimization, Infinity generates a high-quality 1024x1024 image in 0.8 seconds, making it 2.6x faster than SD3-Medium and establishing it as the fastest text-to-image model. Models and codes will be released to promote further exploration of Infinity for visual generation and unified tokenizer modeling.

cs.CV↗

arXiv2025

Are classical deep neural networks weakly adversarially robust?

Adversarial attacks have received increasing attention and it has been widely recognized that classical DNNs have weak adversarial robustness. The most commonly used adversarial defense method, adversarial training, improves the adversarial accuracy of DNNs by generating adversarial examples and retraining the model. However, adversarial training requires a significant computational overhead. In this paper, inspired by existing studies focusing on the clustering properties of DNN output features at each layer and the Progressive Feedforward Collapse phenomenon, we propose a method for adversarial example detection and image recognition that uses layer-wise features to construct feature paths and computes the correlation between the examples feature paths and the class-centered feature paths. Experimental results show that the recognition method achieves 82.77% clean accuracy and 44.17% adversarial accuracy on the ResNet-20 with PFC. Compared to the adversarial training method with 77.64% clean accuracy and 52.94% adversarial accuracy, our method exhibits a trade-off without relying on computationally expensive defense strategies. Furthermore, on the standard ResNet-18, our method maintains this advantage with respective metrics of 80.01% and 46.1%. This result reveals inherent adversarial robustness in DNNs, challenging the conventional understanding of the weak adversarial robustness in DNNs.

cs.CV↗

arXiv2025

Bridging Generative and Discriminative Learning: Few-Shot Relation Extraction via Two-Stage Knowledge-Guided Pre-training

Few-Shot Relation Extraction (FSRE) remains a challenging task due to the scarcity of annotated data and the limited generalization capabilities of existing models. Although large language models (LLMs) have demonstrated potential in FSRE through in-context learning (ICL), their general-purpose training objectives often result in suboptimal performance for task-specific relation extraction. To overcome these challenges, we propose TKRE (Two-Stage Knowledge-Guided Pre-training for Relation Extraction), a novel framework that synergistically integrates LLMs with traditional relation extraction models, bridging generative and discriminative learning paradigms. TKRE introduces two key innovations: (1) leveraging LLMs to generate explanation-driven knowledge and schema-constrained synthetic data, addressing the issue of data scarcity; and (2) a two-stage pre-training strategy combining Masked Span Language Modeling (MSLM) and Span-Level Contrastive Learning (SCL) to enhance relational reasoning and generalization. Together, these components enable TKRE to effectively tackle FSRE tasks. Comprehensive experiments on benchmark datasets demonstrate the efficacy of TKRE, achieving new state-of-the-art performance in FSRE and underscoring its potential for broader application in low-resource scenarios. \footnote{The code and data are released on https://github.com/UESTC-GQJ/TKRE.

cs.CL↗

arXiv2025

Lam-Tung relation breaking in $Z$ boson production as a probe of SMEFT effects

The violation of Lam-Tung relation in the high-$p_T^{\ell\ell}$ region of the Drell-Yan process at the LHC presents a long-standing discrepancy with the standard model prediction at $\mathcal{O}(α_s^3)$ accuracy. In this Letter, we employed a model-independent analysis to investigate this phenomenon within the framework of the Standard Model Effective Field Theory (SMEFT). Our findings revealed that the leading contributions from SMEFT to this violation appear at the $1/Λ^4$ order with $\mathcal{O}(α_s)$ accuracy in QCD interaction. Notably, we demonstrated that the quadratic effect of dimension-6 dipole operators, associated with the $Z$ boson, dominates the breaking effects induced by various dimension-6 and dimension-8 operators. This provides a possible explanation for the observed discrepancy with the Standard Model predictions at the LHC. Furthermore, the breaking effects could also serve as a powerful tool for constraining $Z$-boson dipole interactions, highlighting their importance among potential sources of new physics in the Drell-Yan process.

hep-ph↗

arXiv2025

Lam-Tung relation breaking effects and weak dipole moments at lepton colliders

The breaking of the Lam-Tung relation in the Drell-Yan process at the LHC exhibits a long-standing tension with the Standard Model (SM) prediction at $\mathcal{O}(α_s^3)$ accuracy. This tension could be explained by weak dipole interactions of leptons and quarks, associated with the $Z$-boson within the framework of the Standard Model Effective Field Theory (SMEFT). In this paper, we propose to cross-check these weak dipole interactions by measuring the violation effects of the Lam-Tung relation at future lepton colliders through the processes $e^+e^- \to Zγ\to \ell\bar{\ell}γ$ and $e^+e^- \to Zγ\to q\bar{q}γ$. By considering different decay modes of the $Z$-boson, these channels exhibit distinct sensitivities to various dipole operators, providing a way to disentangle their individual effects. Additionally, the high flavor-tagging efficiencies at lepton colliders could provide strong constraints on the dipole interactions of heavy quarks, such as $b$ and $c$ quarks, which are challenging to probe in the Drell-Yan process at the LHC due to the suppression of parton distribution functions.

hep-ph↗

arXiv2025

Consensus About Classical Reality in a Quantum Universe

Quantum Darwinism recognizes that decoherence imprints redundant records of preferred quasi-classical pointer states on the environment. These redundant records are then accessed by observers. We show how redundancy enables and even implies consensus between observers who use fragments of that decohering environment to acquire information about systems of interest. We quantify consensus using information-theoretic measures that employ mutual information to assess the correlation between the records available to observers from distinct -- hence, independently accessible -- fragments of the environment. We prove that when these fragments have enough information about a system, observers that access them will attribute the same pointer state to that system. Thus, those who know enough about the system agree about what they know. We then test proposed measures of consensus in a solvable model of decoherence as well as in numerical simulations of a many-body system. These results provide detailed understanding of how our classical everyday world arises from within the fundamentally quantum Universe we inhabit.

quant-ph↗

arXiv2025

Unlocking the power of global quantum gates with machine learning

In conventional circuit-based quantum computing architectures, the standard gate set includes arbitrary single-qubit rotations and two-qubit entangling gates. This choice is not always aligned with the native operations available in certain hardware, where the natural entangling gates are not restricted to two qubits but can act on multiple, or even all, qubits simultaneously. However, leveraging the capabilities of global quantum operations for algorithm implementations is highly challenging, as directly compiling local gate sequences into global gates usually gives rise to a quantum circuit that is more complex than the original one. Here, we circumvent this difficulty using a variational approach. Specifically, we study parameterized circuit ansatze composed of a finite number of global gates and layers of single-qubit unitaries. We demonstrate the expressibility of these ansatze and apply them to the problem of ground state preparation for the Heisenberg model and the toric code Hamiltonian, highlighting their potential for offering practical advantages.

quant-ph↗