arXiv subjects

Jie Yang

Publications and source records attributed to Jie Yang.

At least 145 records · Page 8Linked to original sources

arXiv2026

EmbeddingRWKV: State-Centric Retrieval with Reusable States

Current Retrieval-Augmented Generation (RAG) systems typically employ a traditional two-stage pipeline: an embedding model for initial retrieval followed by a reranker for refinement. However, this paradigm suffers from significant inefficiency due to the lack of shared information between stages, leading to substantial redundant computation. To address this limitation, we propose \textbf{State-Centric Retrieval}, a unified retrieval paradigm that utilizes "states" as a bridge to connect embedding models and rerankers. First, we perform state representation learning by fine-tuning an RWKV-based LLM, transforming it into \textbf{EmbeddingRWKV}, a unified model that serves as both an embedding model and a state backbone for extracting compact, reusable states. Building upon these reusable states, we further design a state-based reranker to fully leverage precomputed information. During reranking, the model processes only query tokens, decoupling inference cost from document length and yielding a 5.4$\times$--44.8$\times$ speedup. Furthermore, we observe that retaining all intermediate layer states is unnecessary; with a uniform layer selection strategy, our model maintains 98.62\% of full-model performance using only 25\% of the layers. Extensive experiments demonstrate that State-Centric Retrieval achieves high-quality retrieval and reranking results while significantly enhancing overall system efficiency. Code is available at \href{https://github.com/howard-hou/EmbeddingRWKV}{our GitHub repository}.

cs.CL↗

arXiv2026

Fully Coupled Multiphysics Modelling of Fracture Behaviour in Silicon Particles During Lithiation Delithiation Using the Phase Field Method

In this study, a multiphysics model fully coupling mass transport, deformation, phase field, and fatigue damage was developed to investigate the cracking and fracturing behaviours of Si particles during the single lithiation-delithiation cycle and fatigue damage during multiple cycles. The effects of particle diameter, charge rate, and pre-existing notches on the failure behaviour of Si particles were systematically analysed. The results showed that the increase in charge rate, particle diameter, and pre-existing notch length leads to larger cracking rates and faster fracturing of the particle. Then, a validated contour map of Si particle's fracture behaviours was developed. Additionally, the influence of pre-existing notch length and charge rate on fatigue damage was examined, and it was found that longer pre-existing notch length and larger charge rate can shorten the particle's cyclic life. Finally, to alleviate the particle fracture, nanopores were introduced in the particle, and the influence of porosity on the fracture behaviours was investigated. The results showed that nanopores can reduce expansion, dissipate global tensile stresses and elongate the crack propagation path. The developed computational framework established a predictive relationship between stress diffusion coupling theory and particle-level degradation, guiding future design and manufacturing of failure-resistant Si-based anodes for lithium-ion batteries.

physics.app-ph↗

arXiv2025

Grad: Guided Relation Diffusion Generation for Graph Augmentation in Graph Fraud Detection

Nowadays, Graph Fraud Detection (GFD) in financial scenarios has become an urgent research topic to protect online payment security. However, as organized crime groups are becoming more professional in real-world scenarios, fraudsters are employing more sophisticated camouflage strategies. Specifically, fraudsters disguise themselves by mimicking the behavioral data collected by platforms, ensuring that their key characteristics are consistent with those of benign users to a high degree, which we call Adaptive Camouflage. Consequently, this narrows the differences in behavioral traits between them and benign users within the platform's database, thereby making current GFD models lose efficiency. To address this problem, we propose a relation diffusion-based graph augmentation model Grad. In detail, Grad leverages a supervised graph contrastive learning module to enhance the fraud-benign difference and employs a guided relation diffusion generator to generate auxiliary homophilic relations from scratch. Based on these, weak fraudulent signals would be enhanced during the aggregation process, thus being obvious enough to be captured. Extensive experiments have been conducted on two real-world datasets provided by WeChat Pay, one of the largest online payment platforms with billions of users, and three public datasets. The results show that our proposed model Grad outperforms SOTA methods in both various scenarios, achieving at most 11.10% and 43.95% increases in AUC and AP, respectively. Our code is released at https://github.com/AI4Risk/antifraud and https://github.com/Muyiiiii/WWW25-Grad.

cs.LG↗

arXiv2025

Topological flatness of orthogonal spin local models

Let $p$ be an odd prime and $F$ be a complete discretely valued field with residue field of characteristic $p$. For any parahoric level structure of the split even orthogonal similitude group $\operatorname{GO}_{2n}$ over $F$, we prove a preliminary form of the Pappas-Rapoport flatness conjecture: the associated spin local model is topologically flat.

math.NT↗

arXiv2025

Insight Miner: A Time Series Analysis Dataset for Cross-Domain Alignment with Natural Language

Time-series data is critical across many scientific and industrial domains, including environmental analysis, agriculture, transportation, and finance. However, mining insights from this data typically requires deep domain expertise, a process that is both time-consuming and labor-intensive. In this paper, we propose \textbf{Insight Miner}, a large-scale multimodal model (LMM) designed to generate high-quality, comprehensive time-series descriptions enriched with domain-specific knowledge. To facilitate this, we introduce \textbf{TS-Insights}\footnote{Available at \href{https://huggingface.co/datasets/zhykoties/time-series-language-alignment}{https://huggingface.co/datasets/zhykoties/time-series-language-alignment}.}, the first general-domain dataset for time series and language alignment. TS-Insights contains 100k time-series windows sampled from 20 forecasting datasets. We construct this dataset using a novel \textbf{agentic workflow}, where we use statistical tools to extract features from raw time series before synthesizing them into coherent trend descriptions with GPT-4. Following instruction tuning on TS-Insights, Insight Miner outperforms state-of-the-art multimodal models, such as LLaVA \citep{liu2023llava} and GPT-4, in generating time-series descriptions and insights. Our findings suggest a promising direction for leveraging LMMs in time series analysis, and serve as a foundational step toward enabling LLMs to interpret time series as a native input modality.

cs.LG↗

arXiv2025

Spatial and temporal study of the post-compressed high-power laser pulses for coherent extreme ultraviolet source development

We compared the performance of two post-compression techniques, a gas-filled hollow-core fiber (HCF) and a multi-pass cell (MPC), using a high-power ytterbium-doped fiber laser. The HCF produced 27 fs pulses from 230 fs inputs at >50% efficiency, whereas the MPC achieved 34 fs pulses with significantly higher efficiency (>88%). Both results aligned well with numerical simulations. Crucially, spatial wavefront analysis revealed that the HCF acts as a modal filter, improving beam quality, whereas the MPC introduces aberrations through cumulative mirror errors. Furthermore, we characterize the photon flux of high harmonic generation driven by the post-compressed pulses from the HCF and MPC. These finding highlights that post-compression technique based on self-phase modulation is efficient for the intensity boosting of femtosecond laser system, providing opportunities for generating high quality extreme ultraviolet (XUV) sources. In addition, further improvement of spatial wavefront quality is suggested using the HCF as a single compressor or output component of the cascade compressor.

physics.optics↗

arXiv2025

Why Chain of Thought Fails in Clinical Text Understanding

Large language models (LLMs) are increasingly being applied to clinical care, a domain where both accuracy and transparent reasoning are critical for safe and trustworthy deployment. Chain-of-thought (CoT) prompting, which elicits step-by-step reasoning, has demonstrated improvements in performance and interpretability across a wide range of tasks. However, its effectiveness in clinical contexts remains largely unexplored, particularly in the context of electronic health records (EHRs), the primary source of clinical documentation, which are often lengthy, fragmented, and noisy. In this work, we present the first large-scale systematic study of CoT for clinical text understanding. We assess 95 advanced LLMs on 87 real-world clinical text tasks, covering 9 languages and 8 task types. Contrary to prior findings in other domains, we observe that 86.3\% of models suffer consistent performance degradation in the CoT setting. More capable models remain relatively robust, while weaker ones suffer substantial declines. To better characterize these effects, we perform fine-grained analyses of reasoning length, medical concept alignment, and error profiles, leveraging both LLM-as-a-judge evaluation and clinical expert evaluation. Our results uncover systematic patterns in when and why CoT fails in clinical contexts, which highlight a critical paradox: CoT enhances interpretability but may undermine reliability in clinical text tasks. This work provides an empirical basis for clinical reasoning strategies of LLMs, highlighting the need for transparent and trustworthy approaches.

cs.CL↗

arXiv2025

Bruxism Recognition via Wireless Signal

Bruxism is an oromandibular movement disorder involving teeth grinding and clenching, which severely impairs sleep quality and dental health. However, its diagnosis remains challenging, as existing methods often cause discomfort or compromise user privacy. To address these limitations, we establish a contactless bruxism recognition system based on millimeter-wave radar. First, we analyzed the potential impact of the movement patterns of teeth grinding on radar echo signals. Based on this analysis, 11 features were extracted. Subsequently, using these features, we performed classification with a Random Forest model on the dataset constructed via millimeter-wave radar. Experimental results demonstrate that the proposed method achieves an accuracy of 96.1% on the test set, with precision, recall, and F1-score all remaining at a relatively high level. This study validates the effectiveness of millimeter-wave radar for SB recognition, providing a non-invasive and privacy-friendly alternative to existing recognition techniques. Future research will focus on enhancing the robustness of the method across diverse populations and environments, as well as striving to mitigate the interference of other facial micro-movements on teeth grinding recognition.

eess.SP↗

arXiv2025

Nex-N1: Agentic Models Trained via a Unified Ecosystem for Large-Scale Environment Construction

The evolution of Large Language Models (LLMs) from passive responders to autonomous agents necessitates a fundamental shift in learning paradigms -- from static imitation to incentive-driven decision making. However, this transition is significantly impeded by the lack of scalable infrastructure capable of constructing high-quality interaction signals for effective policy learning. To address this, we introduce a comprehensive method designed to systematically scale the diversity and complexity of interactive environments. Our method realizes this scaling by addressing three orthogonal dimensions: (1) Complexity: NexAU, a flexible agent framework that supports building complex agent hierarchies via simple configurations; (2) Diversity: NexA4A automatically generates diverse agent hierarchies from natural language to cover infinite domains; and (3) Fidelity: NexGAP bridges the simulation-reality gap by integrating dynamic real-world environment for grounded trajectories synthesis. We train Nex-N1 upon the diverse and complex interactive environments established by our infrastructure. Empirical results on benchmarks such as SWE-bench and tau2 demonstrate that Nex-N1 consistently outperforms SOTA open-source models and achieves competitive performance against frontier proprietary models on complex agentic tasks. We open-source the Nex ecosystem and model weights to facilitate further research.

cs.CL↗

arXiv2025

The diverse morphology of gravitational wave signals from merging neutron-star white-dwarf binaries

In sufficiently compact neutron star-white dwarf (NSWD) binary systems, orbital decay means the white dwarf eventually fills its shrinking Roche lobe, initiating a phase of mass transfer. The exchange of angular momentum-both internal and external-plays a critical role in determining the binary's evolutionary outcome. For neutron stars with relatively low magnetic fields and spin frequencies, whether the orbital separation continues to shrink depends on the interplay between gravitational wave (GW) radiation and mass transfer dynamics. We compute the orbital evolution of NSWD binaries across a broad parameter space, incorporating four key variables. Our results reveal distinct boundaries in the NS-WD mass-mass diagram: binaries with white dwarf masses above these thresholds undergo rapid orbital decay and direct coalescence. The dependence of these boundaries on system parameters indicates that Roche-lobe-filling NSWD binaries can follow multiple evolutionary pathways -- a phenomenon we refer to as branched or polymorphic evolution. NSWD binary systems emit strong and diverse GW signals, many of which would be detectable by space-based GW observatories. The morphology of the evolving GW waveform provides a direct diagnostic for the NSWD binary configuration, including any contribution from an accretion disk. Our models can provide critical waveform templates for identifying merging binary signals in real-time GW data.

astro-ph.HE↗

arXiv2025

Super-Resolution ISAC Receivers: An MCMC-Based Gridless Sparse Bayesian Learning Approach

Integrated sensing and communication (ISAC) is crucial for low-altitude wireless networks (LAWNs), where the safety-critical demand for high-accuracy sensing creates a trade-off between precision and complexity for conventional methods. To address this, we propose a novel gridless sparse Bayesian learning (SBL) framework for joint super-resolution multi-target detection and high-accuracy parameter estimation with manageable computational cost. Our model treats target parameters as continuous variables to bypass the grid limitations of conventional approaches. This SBL formulation, however, transforms the estimation task into a challenging high-dimensional inference problem, which we address by developing an enhanced gradient-based Markov chain Monte Carlo algorithm. Our method integrates mini-batch sampling and the Adam optimizer to ensure computational efficiency and rapid convergence. Finally, we validate the framework's robustness in strong clutter and provide a theoretical benchmark by deriving the corresponding Bayesian Cramer-Rao bound. Simulation results demonstrate remarkable super-resolution capabilities, successfully resolving multiple targets separated by merely 50% of the Rayleigh limit in range, 17% in velocity, and 52% in angle. At a signal-to-noise ratio of 20 dB, the algorithm achieves a multi-target detection probability exceeding 90% while concurrently delivering ultra-high accuracy, with root mean square error of 0.07 m, 0.024 m/s, and 0.015 degree for range, velocity, and angle, respectively. This robust performance, demonstrated against strong clutter, showcases its suitability for practical ISAC-LAWNs applications.

eess.SP↗

arXiv2025

Terahertz Channel performance in ULEO Satellite-to-Ground Communications

The exponential growth in satellite data traffic demands communication systems exceeding current microwave capacity limitations, while the terahertz (THz) frequency band (0.1-10 THz) offers unprecedented bandwidth potential with superior weather resilience compared to optical systems, particularly when combined with ultra low Earth orbit (ULEO) satellite deployments below 300 km altitude. This article presents a comprehensive performance evaluation for ULEO-THz satellite to ground communications, analyzing three distinct transmission architectures direct satellite to ground (S2G), satellite relay ground (SRG) forwarding, and satellite-to-high altitude base station (S2H) with fiber backhaul. Our analysis leverages altitude-resolved atmospheric propagation models validated using year long meteorological data from four high-altitude stations in Tibet and Qinghai, China. It incorporates frequency-dependent atmospheric absorption using ITU-R standards, free space path loss with curved atmospheric modeling, and regional atmospheric variations to derive total channel path loss, available bandwidth capacity, and bit error rate (BER) performance under both AWGN and Weibull fading conditions across multiple THz frequencies. Results demonstrate that direct S2G transmission at lower THz frequencies achieves optimal practical performance with maximum available bandwidth under QPSK modulation, while SRG suffers prohibitive cumulative losses from multiple hops, and S2H is rendered impractical for long-haul links by substantial electro-optical conversion and fiber transmission losses.

physics.app-ph↗

arXiv2025

SMoFi: Step-wise Momentum Fusion for Split Federated Learning on Heterogeneous Data

Split Federated Learning is a system-efficient federated learning paradigm that leverages the rich computing resources at a central server to train model partitions. Data heterogeneity across silos, however, presents a major challenge undermining the convergence speed and accuracy of the global model. This paper introduces Step-wise Momentum Fusion (SMoFi), an effective and lightweight framework that counteracts gradient divergence arising from data heterogeneity by synchronizing the momentum buffers across server-side optimizers. To control gradient divergence over the training process, we design a staleness-aware alignment mechanism that imposes constraints on gradient updates of the server-side submodel at each optimization step. Extensive validations on multiple real-world datasets show that SMoFi consistently improves global model accuracy (up to 7.1%) and convergence speed (up to 10.25$\times$). Furthermore, SMoFi has a greater impact with more clients involved and deeper learning models, making it particularly suitable for model training in resource-constrained contexts.

cs.LG↗

arXiv2025

Impact of mass transfer on the orbital evolution of a white dwarf close to an intermediate-mass black hole

Extreme mass ratio inspirals (EMRIs) of low-mass white dwarfs (WDs, 0.1 - 0.3 Msun) around spinning intermediate-mass black holes (IMBHs, 10^3 - 10^5 Msun) offer unique opportunities for multi-messenger astronomy, emitting both gravitational waves (GWs) and electromagnetic (EM) signals. Yet, despite their astrophysical relevance, theoretical models often omit key interactions between relativistic dynamics and phase-dependent mass transfer (MT). In this study, we integrate a perturbed Keplerian formalism with post-Newtonian (PN) corrections to simulate the relativistic orbit of a WD around a rotating IMBH, explicitly resolving the narrow phase near pericentre where Roche-lobe overflow initiates MT. We find that GW emission and MT exert competing influences on the orbit: MT episodes can increase both orbital period and eccentricity, potentially enabling the WD to avoid complete tidal disruption and even escape. We further quantify the GW phase evolution induced by MT, identifying parameter regimes in which GW detectors could observe a one-radian phase shift over observational timescales. Finally, we propose that the orbital expansion driven by MT may lead to the disappearance of quasi-periodic eruptions (QPEs). Our results underscore the necessity of jointly modeling relativistic effects and dynamic mass transfer in WD-IMBH systems.

astro-ph.HE↗

arXiv2025

Scalable Medication Extraction and Discontinuation Identification from Electronic Health Records Using Large Language Models

Identifying medication discontinuations in electronic health records (EHRs) is vital for patient safety but is often hindered by information being buried in unstructured notes. This study aims to evaluate the capabilities of advanced open-sourced and proprietary large language models (LLMs) in extracting medications and classifying their medication status from EHR notes, focusing on their scalability on medication information extraction without human annotation. We collected three EHR datasets from diverse sources to build the evaluation benchmark. We evaluated 12 advanced LLMs and explored multiple LLM prompting strategies. Performance on medication extraction, medication status classification, and their joint task (extraction then classification) was systematically compared across all experiments. We found that LLMs showed promising performance on the medication extraction and discontinuation classification from EHR notes. GPT-4o consistently achieved the highest average F1 scores in all tasks under zero-shot setting - 94.0% for medication extraction, 78.1% for discontinuation classification, and 72.7% for the joint task. Open-sourced models followed closely, Llama-3.1-70B-Instruct achieved the highest performance in medication status classification on the MIV-Med dataset (68.7%) and in the joint task on both the Re-CASI (76.2%) and MIV-Med (60.2%) datasets. Medical-specific LLMs demonstrated lower performance compared to advanced general-domain LLMs. Few-shot learning generally improved performance, while CoT reasoning showed inconsistent gains. LLMs demonstrate strong potential for medication extraction and discontinuation identification on EHR notes, with open-sourced models offering scalable alternatives to proprietary systems and few-shot can further improve LLMs' capability.

cs.CL↗

arXiv2025

NIR-II Fluorescence Project Technology for Augmented Reality Surgical Navigation

NIR-II fluorescence imaging provides superior tissue penetration and clarity, yet its clinical use in surgical navigation is hindered by a critical workflow issue. Surgeons must divert their attention between the operative field and external monitors, increasing cognitive load and disrupting procedures. Current strategies have failed to resolve this fundamental problem. Here, we developed a co-axial NIR-II fluorescence projection navigation system to enable real-time, in situ visualization. This system creates an intraoperative augmented reality by directly projecting high-precision, pseudocolored fluorescence images onto the surgical field, spatially integrating functional signals with patient anatomy. Validated through in vitro, in vivo, and clinical patient studies, our system eliminates visual field switching, reduces intraoperative distraction, and preserves natural stereoscopic vision. This approach represents a paradigm shift toward a more coherent, efficient, and ergonomically optimized optical imaging modality for surgical navigation.

physics.optics↗

arXiv2025

A Memory-Efficient Retrieval Architecture for RAG-Enabled Wearable Medical LLMs-Agents

With powerful and integrative large language models (LLMs), medical AI agents have demonstrated unique advantages in providing personalized medical consultations, continuous health monitoring, and precise treatment plans. Retrieval-Augmented Generation (RAG) integrates personal medical documents into LLMs by an external retrievable database to address the costly retraining or fine-tuning issues in deploying customized agents. While deploying medical agents in edge devices ensures privacy protection, RAG implementations impose substantial memory access and energy consumption during the retrieval stage. This paper presents a hierarchical retrieval architecture for edge RAG, leveraging a two-stage retrieval scheme that combines approximate retrieval for candidate set generation, followed by high-precision retrieval on pre-selected document embeddings. The proposed architecture significantly reduces energy consumption and external memory access while maintaining retrieval accuracy. Simulation results show that, under TSMC 28nm technology, the proposed hierarchical retrieval architecture has reduced the overall memory access by nearly 50% and the computation by 75% compared to pure INT8 retrieval, and the total energy consumption for 1 MB data retrieval is 177.76 μJ/query.

cs.AR↗

arXiv2025

Fast Penning Ionization of Cold Rydberg atoms in an Electric Field

We observe a fast Penning ionization in a dilute gas of cold rubidium Rydberg atoms, in the presence of a static electric field of 50 V/cm, with the ionization rate coefficients for two specific states being measured, which are orders of magnitude higher than the theoretical predictions in field-free space. Our analysis based on a polarized two-atom model reveals that the ionization threshold of Rydberg atoms is lowered by the static electric field, reducing the energy exchange required for Penning ionization and increasing the ionization rate. Beyond this, the dipole-dipole interaction strengthened by the electric field between two Rydberg atoms at a micrometer-scale distance leads to double ionization of the atoms pair, opening a new autoionization channel. Such enhancement of the Penning ionization by a static electric field poses both a threat to the stability and a potential control strategy for quantum systems composed of cold Rydberg atoms with micrometer-scale interatomic separations.

physics.atom-ph↗