arXiv subjects

Wen Luo

Publications and source records attributed to Wen Luo.

At least 37 records · Page 2Linked to original sources

arXiv2025

Ear-Keeper: A Cross-Platform AI System for Rapid and Accurate Ear Disease Diagnosis

Early and accurate detection systems for ear diseases, powered by deep learning, are essential for preventing hearing impairment and improving population health. However, the limited diversity of existing otoendoscopy datasets and the poor balance between diagnostic accuracy, computational efficiency, and model size have hindered the translation of artificial intelligence (AI) algorithms into healthcare applications. In this study, we constructed a large-scale, multi-center otoendoscopy dataset covering eight common ear diseases and healthy cases. Building upon this resource, we developed Best-EarNet, an ultrafast and lightweight deep learning architecture integrating a novel Local-Global Spatial Feature Fusion Module with a multi-scale supervision strategy, enabling real-time and accurate classification of ear conditions. Leveraging transfer learning, Best-EarNet, with a model size of only 2.94 MB, achieved diagnostic accuracies of 95.23% on an internal test set (22,581 images) and 92.14% on an external test set (1,652 images), while requiring only 0.0125 seconds (80 frames per second) to process a single image on a standard CPU. Further subgroup analysis by gender and age showed consistently excellent performance of Best-EarNet across all demographic groups. To enhance clinical interpretability and user trust, we incorporated Grad-CAM-based visualization, highlighting the specific abnormal ear regions contributing to AI predictions. Most importantly, we developed Ear-Keeper, a cross-platform intelligent diagnosis system built upon Best-EarNet, deployable on smartphones, tablets, and personal computers. Ear-Keeper enables public users and healthcare providers to perform comprehensive real-time video-based ear canal screening, supporting early detection and timely intervention of ear diseases.

cs.CV↗

arXiv2025

FEA-Bench: A Benchmark for Evaluating Repository-Level Code Generation for Feature Implementation

Implementing new features in repository-level codebases is a crucial application of code generation models. However, current benchmarks lack a dedicated evaluation framework for this capability. To fill this gap, we introduce FEA-Bench, a benchmark designed to assess the ability of large language models (LLMs) to perform incremental development within code repositories. We collect pull requests from 83 GitHub repositories and use rule-based and intent-based filtering to construct task instances focused on new feature development. Each task instance containing code changes is paired with relevant unit test files to ensure that the solution can be verified. The feature implementation requires LLMs to simultaneously possess code completion capabilities for new components and code editing abilities for other relevant parts in the code repository, providing a more comprehensive evaluation method of LLMs' automated software engineering capabilities. Experimental results show that LLMs perform significantly worse in the FEA-Bench, highlighting considerable challenges in such repository-level incremental code development.

cs.SE↗

arXiv2025

Well Begun is Half Done: Low-resource Preference Alignment by Weak-to-Strong Decoding

Large Language Models (LLMs) require alignment with human preferences to avoid generating offensive, false, or meaningless content. Recently, low-resource methods for LLM alignment have been popular, while still facing challenges in obtaining both high-quality and aligned content. Motivated by the observation that the difficulty of generating aligned responses is concentrated at the beginning of decoding, we propose a novel framework, Weak-to-Strong Decoding (WSD), to enhance the alignment ability of base models by the guidance of a small aligned model. The small model first drafts well-aligned beginnings, followed by the large base model to continue the rest, controlled by a well-designed auto-switch mechanism. We also collect a new dataset, GenerAlign, to fine-tune a small-sized Pilot-3B as the draft model, which effectively enhances different base models under the WSD framework to outperform all baseline methods, while avoiding degradation on downstream tasks, termed as the alignment tax. Extensive experiments are further conducted to examine the impact of different settings and time efficiency, as well as analyses on the intrinsic mechanisms of WSD in depth.

cs.CL↗

arXiv2025

Odysseus Navigates the Sirens' Song: Dynamic Focus Decoding for Factual and Diverse Open-Ended Text Generation

Large Language Models (LLMs) are increasingly required to generate text that is both factually accurate and diverse across various open-ended applications. However, current stochastic decoding methods struggle to balance such objectives. We introduce Dynamic Focus Decoding (DFD), a novel plug-and-play stochastic approach that resolves this trade-off without requiring additional data, knowledge, or models. DFD adaptively adjusts the decoding focus based on distributional differences across layers, leveraging the modular and hierarchical nature of factual knowledge within LLMs. This dynamic adjustment improves factuality in knowledge-intensive decoding steps and promotes diversity in less knowledge-reliant steps. DFD can be easily integrated with existing decoding methods, enhancing both factuality and diversity with minimal computational overhead. Extensive experiments across seven datasets demonstrate that DFD significantly improves performance, providing a scalable and efficient solution for open-ended text generation.

cs.CL↗

arXiv2025

GeoRAG: A Question-Answering Approach from a Geographical Perspective

Geographic Question Answering (GeoQA) addresses natural language queries in geographic domains to fulfill complex user demands and improve information retrieval efficiency. Traditional QA systems, however, suffer from limited comprehension, low retrieval accuracy, weak interactivity, and inadequate handling of complex tasks, hindering precise information acquisition. This study presents GeoRAG, a knowledge-enhanced QA framework integrating domain-specific fine-tuning and prompt engineering with Retrieval-Augmented Generation (RAG) technology to enhance geographic knowledge retrieval accuracy and user interaction. The methodology involves four components: (1) A structured geographic knowledge base constructed from 3267 corpora (research papers, monographs, and technical reports), categorized via a multi-agent approach into seven dimensions: semantic understanding, spatial location, geometric morphology, attribute characteristics, feature relationships, evolutionary processes, and operational mechanisms. This yielded 145234 classified entries and 875432 multi-dimensional QA pairs. (2) A multi-label text classifier based on BERT-Base-Chinese, trained to analyze query types through geographic dimension classification. (3) A retrieval evaluator leveraging QA pair data to assess query-document relevance, optimizing retrieval precision. (4) GeoPrompt templates engineered to dynamically integrate user queries with retrieved information, enhancing response quality through dimension-specific prompting. Comparative experiments demonstrate GeoRAG's superior performance over conventional RAG across multiple base models, validating its generalizability. This work advances geographic AI by proposing a novel paradigm for deploying large language models in domain-specific contexts, with implications for improving GeoQA systems scalability and accuracy in real-world applications.

cs.IR↗

arXiv2025

Explanation based In-Context Demonstrations Retrieval for Multilingual Grammatical Error Correction

Grammatical error correction (GEC) aims to correct grammatical, spelling, and semantic errors in natural language text. With the growing of large language models (LLMs), direct text generation has gradually become the focus of the GEC methods, and few-shot in-context learning presents a cost-effective solution. However, selecting effective in-context examples remains challenging, as the similarity between input texts does not necessarily correspond to similar grammatical error patterns. In this paper, we propose a novel retrieval method based on natural language grammatical error explanations (GEE) to address this issue. Our method retrieves suitable few-shot demonstrations by matching the GEE of the test input with that of pre-constructed database samples, where explanations for erroneous samples are generated by LLMs. We conducted multilingual GEC few-shot experiments on both major open-source and closed-source LLMs. Experiments across five languages show that our method outperforms existing semantic and BM25-based retrieval techniques, without requiring additional training or language adaptation. This also suggests that matching error patterns is key to selecting examples.

cs.CL↗

arXiv2024

SiamSeg: Self-Training with Contrastive Learning for Unsupervised Domain Adaptation Semantic Segmentation in Remote Sensing

Semantic segmentation of remote sensing (RS) images is a challenging yet essential task with broad applications. While deep learning, particularly supervised learning with large-scale labeled datasets, has significantly advanced this field, the acquisition of high-quality labeled data remains costly and time-intensive. Unsupervised domain adaptation (UDA) provides a promising alternative by enabling models to learn from unlabeled target domain data while leveraging labeled source domain data. Recent self-training (ST) approaches employing pseudo-label generation have shown potential in mitigating domain discrepancies. However, the application of ST to RS image segmentation remains underexplored. Factors such as variations in ground sampling distance, imaging equipment, and geographic diversity exacerbate domain shifts, limiting model performance across domains. In that case, existing ST methods, due to significant domain shifts in cross-domain RS images, often underperform. To address these challenges, we propose integrating contrastive learning into UDA, enhancing the model's ability to capture semantic information in the target domain by maximizing the similarity between augmented views of the same image. This additional supervision improves the model's representational capacity and segmentation performance in the target domain. Extensive experiments conducted on RS datasets, including Potsdam, Vaihingen, and LoveDA, demonstrate that our method, SimSeg, outperforms existing approaches, achieving state-of-the-art results. Visualization and quantitative analyses further validate SimSeg's superior ability to learn from the target domain. The code is publicly available at https://github.com/woldier/SiamSeg.

cs.CV↗

arXiv2024

Efficient transmutation of long-lived fission products in a Gamma Factory beam driven advanced nuclear energy system

The Gamma Factory (GF) project aims to generate high-intensity $γ$-ray beams of tunable energy and relatively small energy spread. Such beams can be optimized to generate an intense photo-neutron source, capable of driving an advanced nuclear energy system (ANES) for nuclear waste transmutation and supplying electrical power that is necessary for the GF operation mode of the Large Hadron Collider storage ring. In this study, we investigate the feasibility of driving ANES with the GF beam which is optimized to maximize the neutron production rate. The dependence of the ANES thermal power on the distance between the positions of the ANES and the GF $γ$-ray source is evaluated. For the $γ$-ray beam reaching the intensity of $\sim$$10^{19}$ photons per second, the ANES thermal power could exceed $500\,$MWt. Under the assumption that ANES operates over $20$ years, the transmutation rate could reach $30\%$ for five typical long-lived fission products (LLFPs): $^{79}$Se, $^{99}$Tc, $^{107}$Pd, $^{129}$I, $^{137}$Cs. Our comparative studies show that although the neutron production efficiency of the GF $γ$-ray beam (per MW of the beam power) is approximately $14$ times lower than that of the $500\,$MeV proton beam, the overall net ANES power production efficiency for the GF beam driver scheme could be comparable to that of the proton beam driver scheme, while providing additional transmutation capacity, not available for the proton beam driven scheme. It is suggested that the GF-based ANES could provide a viable solution for the efficient transmutation of LLFPs without isotopic separation.

physics.acc-ph↗

arXiv2024

Shapley Value-based Contrastive Alignment for Multimodal Information Extraction

The rise of social media and the exponential growth of multimodal communication necessitates advanced techniques for Multimodal Information Extraction (MIE). However, existing methodologies primarily rely on direct Image-Text interactions, a paradigm that often faces significant challenges due to semantic and modality gaps between images and text. In this paper, we introduce a new paradigm of Image-Context-Text interaction, where large multimodal models (LMMs) are utilized to generate descriptive textual context to bridge these gaps. In line with this paradigm, we propose a novel Shapley Value-based Contrastive Alignment (Shap-CA) method, which aligns both context-text and context-image pairs. Shap-CA initially applies the Shapley value concept from cooperative game theory to assess the individual contribution of each element in the set of contexts, texts and images towards total semantic and modality overlaps. Following this quantitative evaluation, a contrastive learning strategy is employed to enhance the interactive contribution within context-text/image pairs, while minimizing the influence across these pairs. Furthermore, we design an adaptive fusion module for selective cross-modal fusion. Extensive experiments across four MIE datasets demonstrate that our method significantly outperforms existing state-of-the-art methods.

cs.AI↗

arXiv2024

HalluDial: A Large-Scale Benchmark for Automatic Dialogue-Level Hallucination Evaluation

Large Language Models (LLMs) have significantly advanced the field of Natural Language Processing (NLP), achieving remarkable performance across diverse tasks and enabling widespread real-world applications. However, LLMs are prone to hallucination, generating content that either conflicts with established knowledge or is unfaithful to the original sources. Existing hallucination benchmarks primarily focus on sentence- or passage-level hallucination detection, neglecting dialogue-level evaluation, hallucination localization, and rationale provision. They also predominantly target factuality hallucinations while underestimating faithfulness hallucinations, often relying on labor-intensive or non-specialized evaluators. To address these limitations, we propose HalluDial, the first comprehensive large-scale benchmark for automatic dialogue-level hallucination evaluation. HalluDial encompasses both spontaneous and induced hallucination scenarios, covering factuality and faithfulness hallucinations. The benchmark includes 4,094 dialogues with a total of 146,856 samples. Leveraging HalluDial, we conduct a comprehensive meta-evaluation of LLMs' hallucination evaluation capabilities in information-seeking dialogues and introduce a specialized judge language model, HalluJudge. The high data quality of HalluDial enables HalluJudge to achieve superior or competitive performance in hallucination evaluation, facilitating the automatic assessment of dialogue-level hallucinations in LLMs and providing valuable insights into this phenomenon. The dataset and the code are available at https://github.com/FlagOpen/HalluDial.

cs.CL↗

arXiv2023

Efficient production of nuclear isomer $^{93m}$Mo with laser-accelerated proton beam and an astrophysical implication on $^{92m}$Mo production

Nuclear isomers play a key role in the creation of the elements in the universe and have a number of fascinating potential applications related to the controlled release of nuclear energy on demand. Particularly, $^{93m}$Mo isomer is a good candidate for studying the depletion of nuclear isomer via nuclear excitation by electron capture. For such purposes, efficient approach for $^{93m}$Mo production needs to be explored. In the present work, we demonstrate experimentally an efficient production of $^{93m}$Mo through $^{93}$Nb(p, n) reaction induced by intense laser pulse. When a ps-duration, 100-J laser pulse is employed, the $^{93m}$Mo isomer at 2425 keV (21/2$^+$, $T_{1/2}$ = 6.85 h) are generated with a high yield of $1.8\times10^6$ particles/shot. The resulting peak efficiency is expected to be $10^{17}$ particles/s, which is at least five orders of magnitudes higher than using classical proton accelerator. The effects of production and destruction of $^{93m}$Mo on the controversial astrophysical p-isotope $^{92}$Mo are studied. It is found that the $^{93}$Nb(p, n)-$^{93m}$Mo reaction is an important production path for ^{93m}Mo seed nucleus, and the influence of ^{93m}Mo-^{92}Mo reaction flow on ^{92}Mo production cannot be ignored. In addition, we propose to directly measure the astrophysical rate of (p, n) reaction using laser-induced proton beam since the latter one fits the Maxwell-Boltzmann distribution well. We conclude that laser-induced proton beam opens a new path to produce nuclear isomers with high peak efficiency towards the understanding of p-nuclei nucleosythesis.

nucl-ex↗

arXiv2022

High-energy-density plasma in femtosecond-laser-irradiated nanowire array targets for nuclear reactions

In this work, the high-energy-density plasmas (HEDP) evolved from joule-class-femtosecond-laser-irradiated nanowire array (NWA) targets are numerically and experimentally studied. The particle-in-cell (PIC) simulations indicate that ions accelerated in the sheath field around the nanowires' surface were eventually confined in NWA plasma, contributing most to the high energy densities. The protons emitted from the front surface of targets provide rich information about the interaction. The electron and ion energy densities in a broad target parameter range are given. Compared to planar targets, the ion energy density is one order of magnitude higher, and the volume of the HEDP is several-fold larger. At optimal target parameters, 8% of the laser energy can be converted to confined protons and results in ion energy densities of up to GJ/cm3 level. Experimental measurements of the emitted ions and neutrons from 2H(d, n)3He fusion from polyethylene and deuterated polyethylene NWA targets confirm the above results.

physics.plasm-ph↗

arXiv2022

Alpha-particle generation from H-11B fusion initiated by laser-accelerated boron ions

Here we report the generation of MeV alpha-particles from H-11B fusion initiated by laser-accelerated boron ions. Boron ions with maximum energy of 6MeV and fluence of 10^9/MeV/sr@5MeV were generated from 60-nm-thick self-supporting boron nanofoils irradiated by 1J femtosecond pulses at an intensity of 10^19W/cm^2. By bombarding secondary hydrogenous targets with the boron ions, 3*10^5/sr alpha-particles from H-11B fusion were registered, which is consistent with the theoretical yield calculated from the measured boron energy spectra. Our results demonstrate an alternative way toward ultrashort MeV alpha-particle sources employing compact femtosecond lasers. The ion acceleration and product measurement scheme are referential for the studies on the ion stopping power and cross-section of the H-11B reaction in solid or plasma.

physics.plasm-ph↗

arXiv2021

Rapid interrogation of special nuclear materials by combining scattering and transmission nuclear resonance fluorescence spectroscopy

The smuggling of special nuclear materials (SNMs) across national borders is becoming a serious threat to nuclear nonproliferation. This paper presents a feasibility study on the rapid interrogation of concealed SNMs by combining scattering and transmission nuclear resonance fluorescence (sNRF and tNRF) spectroscopy. In sNRF spectroscopy, SNMs such as $^{235, 238}$U are excited by a wide-band photon beam of appropriate energy and exhibit unique NRF signatures. Monte Carlo simulations show that one-dimensional scans can realize isotopic identification of concealed $^{235, 238}$U when the detector array used for interrogation has sufficiently high energy resolution. The simulated isotopic ratio $^{235}U/^{238}U$ is in good agreement with the theoretical value when the SNMs are enclosed in relatively thin iron. This interrogation is followed by tNRF spectroscopy using a narrow-band photon beam with the goal of obtaining tomographic images of the concealed SNMs. The reconstructed image clearly reveals the position of the isotope $^{235}$U inside an iron rod. It is shown that the interrogation time of sNRF and tNRF spectroscopy is one order of magnitude lower than that when only tNRF spectroscopy is used and results in a missed-detection rate of 10$^{-3}$. The proposed method can also be applied for isotopic imaging of other SNMs such as $^{239, 240}$Pu and $^{237}$Np.

physics.app-ph↗

arXiv2020

Brilliant attosecond γ-ray emission and high-yield positron production from intense laser-irradiated Nano-Micro array

We investigate a novel scheme for brilliant attosecond γ-ray emission and high-yield positron production, which is accomplished with an ultra-intense laser pulse incident upon a Nano-Micro array (NMA) with substrate incorporated. This scheme is able to realize effectively electron acceleration and colliding geometry. Both the γ-ray flash and positron bunch are then generated with high conversion efficiency. At laser intensity of I_0 = 8 times 10^{23} W/cm^2, ~27% of the laser energy is transferred successfully into the γ-rays, and ~0.7% of the laser energy into the positrons. As a consequence, ultra-short (~440 as) and ultra-brilliant (~10^{24} photons s^{-1} mm^{-2} mrad^{-2} per 0.1%BW @ 15 MeV) γ-ray burst, and high-yield (1.48 times 10^{11}) and overdense (~10^{22} cm^{-3}) positron bunch are generated. We found a sub-linear scaling of laser-to-photon conversion efficiency (proportional to I_0^{0.75}) and a super-linear scaling of laser-to-positron conversion efficiency (proportional to I_0^{2.5}) with the laser intensity. Multi-dimensional particle-in-cell simulations show that particle (γ photon and positron) generation can be manipulated by laser-focusing position, and NMA's length and spacing. Optimal conditions for particle generation in NMAs are obtained, indicating that microwire array has the advantage over nanowire array in particle generation in the extreme laser fields. Furthermore, positron annihilation effect in high-energy-density (HED) environment is discussed. The scheme using NMAs would provide effective avenues toward investigating attosecond nuclear science and HED physics with the coming 10 PW laser facilities.

physics.plasm-ph↗

arXiv2020

High-efficiency water-window x-ray generation from nanowire array targets irradiated with femtosecond laser pulses

We demonstrate the high-efficiency generation of water-window soft x-ray emissions from polyethylene nanowire array targets irradiated by femtosecond laser pulses at the intensity of 4*10^19 W/cm^2. The experimental results indicate more than one order of magnitude enhancement of the water-window x-ray emissions from the nanowire array targets compared to the planar targets. The highest energy conversion efficiency from laser to water-window x-rays is measured as 0.5%/sr, which comes from the targets with the longest nanowires. Supported by particle-in-cell simulations and atomic kinetic codes, the physics that leads to the high conversion efficiency is discussed.

physics.plasm-ph↗

arXiv2020

Measurements of D-D fusion neutrons generated in nanowire array laser plasma using Timepix3 detector

We present the results of neutron detection in a laser plasma experiment with a CD$_2$ nanowire target. A hybrid semiconductor pixel detector Timepix3 covered with neutron converters was used for the detection of neutrons. D-D fusion neutrons were detected in a polyethylene converter through recoiled protons. Both the energy of recoiled protons and the time-of-flight of neutrons (and thus their energy) were determined. We report $(2.4 \pm 1.8) \times 10^7$ neutrons generated for 1~J of incoming laser energy. Furthermore, we proved that Timepix3 is suitable for difficult operational conditions in laser experiments.

physics.ins-det↗

arXiv2020

Photo-excitation production of medically interesting isomers using high-intensity γ-ray source

Photon-induced nuclear excitation (i.e. photo-excitation) can be used for production of nuclear isomers, which have potential applications in astrophysics, energy storing, and medical diagnosis and treatment. This paper presents a feasibility study on production of four nuclear isomers ({99m}^Tc, {103m}^Rh and {113m, 115m}^In) using high-intensity γ-ray source based on laser-electron Compton scattering (LCS), for use in the medical diagnosis and treatment. The decay properties and the medical applications of these nuclear isomers were reviewed. The cross-section curves, simulated yields and activity of product of each photo-excitation process were calculated. The cutoff energy of LCS γ-ray beam is optimized by adjusting the electron energy in order to maximize the yields as well as the activities of photo-excitation products. It is found that the achievable activity of above-mentioned isomers can exceed 10 mCi for 6-hour target irradiation at an intensity of the order of 10^{13} γ/s. Such magnitude of activity satisfies the dose requirement of medical diagnosis. Our simulation results suggest the prospect of producing medically interesting isomers with photo-excitation using the state-of-art LCS γ-ray beam facility.

nucl-ex↗