arXiv subjects

Xin Xu

Publications and source records attributed to Xin Xu.

At least 127 records · Page 7Linked to original sources

arXiv2025

Multiple Instance Verification

We explore multiple instance verification, a problem setting in which a query instance is verified against a bag of target instances with heterogeneous, unknown relevancy. We show that naive adaptations of attention-based multiple instance learning (MIL) methods and standard verification methods like Siamese neural networks are unsuitable for this setting: directly combining state-of-the-art (SOTA) MIL methods and Siamese networks is shown to be no better, and sometimes significantly worse, than a simple baseline model. Postulating that this may be caused by the failure of the representation of the target bag to incorporate the query instance, we introduce a new pooling approach named "cross-attention pooling" (CAP). Under the CAP framework, we propose two novel attention functions to address the challenge of distinguishing between highly similar instances in a target bag. Through empirical studies on three different verification tasks, we demonstrate that CAP outperforms adaptations of SOTA MIL methods and the baseline by substantial margins, in terms of both classification accuracy and the ability to detect key instances. The superior ability to identify key instances is attributed to the new attention functions by ablation studies. We share our code at https://github.com/xxweka/MIV.

cs.LG↗

arXiv2025

Conceptual Design Report of Super Tau-Charm Facility: The Accelerator

Electron-positron colliders operating in the GeV region of center-of-mass energies or the Tau-Charm energy region, have been proven to enable competitive frontier research, due to its several unique features. With the progress of high energy physics in the last two decades, a new-generation Tau-Charm factory, Super Tau Charm Facility (STCF) has been actively promoting by the particle physics community in China. STCF holds great potential to address fundamental questions such as the essence of color confinement and the matter-antimatter asymmetry in the universe in the next decades. The main design goals of STCF are with a center-of-mass energy ranging from 2 to 7 GeV and a peak luminosity surpassing 5*10^34 cm^-2s^-1 that is optimized at a center-of-mass energy of 4 GeV, which is about 50 times that of the currently operating Tau-Charm factory - BEPCII. The STCF accelerator is composed of two main parts: a double-ring collider with the crab-waist collision scheme and an injector that provides top-up injections for both electron and positron beams. As a typical third-generation electron-positron circular collider, the STCF accelerator faces many challenges in both accelerator physics and technology. In this paper, the conceptual design of the STCF accelerator complex is presented, including the ongoing efforts and plans for technological R&D, as well as the required infrastructure. The STCF project aims to secure support from the Chinese central government for its construction during the 15th Five-Year Plan (2026-2030) in China.

physics.acc-ph↗

arXiv2025

WenetSpeech-Yue: A Large-scale Cantonese Speech Corpus with Multi-dimensional Annotation

The development of speech understanding and generation has been significantly accelerated by the availability of large-scale, high-quality speech datasets. Among these, ASR and TTS are regarded as the most established and fundamental tasks. However, for Cantonese (Yue Chinese), spoken by approximately 84.9 million native speakers worldwide, limited annotated resources have hindered progress and resulted in suboptimal ASR and TTS performance. To address this challenge, we propose WenetSpeech-Pipe, an integrated pipeline for building large-scale speech corpus with multi-dimensional annotation tailored for speech understanding and generation. It comprises six modules: Audio Collection, Speaker Attributes Annotation, Speech Quality Annotation, Automatic Speech Recognition, Text Postprocessing and Recognizer Output Voting, enabling rich and high-quality annotations. Based on this pipeline, we release WenetSpeech-Yue, the first large-scale Cantonese speech corpus with multi-dimensional annotation for ASR and TTS, covering 21,800 hours across 10 domains with annotations including ASR transcription, text confidence, speaker identity, age, gender, speech quality scores, among other annotations. We also release WSYue-eval, a comprehensive Cantonese benchmark with two components: WSYue-ASR-eval, a manually annotated set for evaluating ASR on short and long utterances, code-switching, and diverse acoustic conditions, and WSYue-TTS-eval, with base and coverage subsets for standard and generalization testing. Experimental results show that models trained on WenetSpeech-Yue achieve competitive results against state-of-the-art (SOTA) Cantonese ASR and TTS systems, including commercial and LLM-based models, highlighting the value of our dataset and pipeline.

cs.SD↗

arXiv2025

CMFDNet: Cross-Mamba and Feature Discovery Network for Polyp Segmentation

Automated colonic polyp segmentation is crucial for assisting doctors in screening of precancerous polyps and diagnosis of colorectal neoplasms. Although existing methods have achieved promising results, polyp segmentation remains hindered by the following limitations,including: (1) significant variation in polyp shapes and sizes, (2) indistinct boundaries between polyps and adjacent tissues, and (3) small-sized polyps are easily overlooked during the segmentation process. Driven by these practical difficulties, an innovative architecture, CMFDNet, is proposed with the CMD module, MSA module, and FD module. The CMD module, serving as an innovative decoder, introduces a cross-scanning method to reduce blurry boundaries. The MSA module adopts a multi-branch parallel structure to enhance the recognition ability for polyps with diverse geometries and scale distributions. The FD module establishes dependencies among all decoder features to alleviate the under-detection of polyps with small-scale features. Experimental results show that CMFDNet outperforms six SOTA methods used for comparison, especially on ETIS and ColonDB datasets, where mDice scores exceed the best SOTA method by 1.83% and 1.55%, respectively.

cs.CV↗

arXiv2025

On Convergence and Stability of Two Extended BB-like Step Sizes

The Barzilai-Borwein (BB) step sizes have a profound impact on gradient descent methods. In this work, we propose two new gradient step sizes: one longer than the original long BB step size, and the other shorter than the original short BB step size. This extends the bounds of the original BB step sizes. For strictly convex quadratic optimization problems, we employ the dynamics of difference equations to prove that these two new methods achieve R-linear convergence. Regarding stability, we surprisingly find that under certain conditions, the gradient descent method corresponding to the longer step size is stable, whereas the shorter step size consistently leads to instability. Numerical results validate these stability theories. Here, stability refers to whether the gradient norm decreases monotonically.

math.OC↗

arXiv2025

Structural-Aware Key Node Identification in Hypergraphs via Representation Learning and Fine-Tuning

Evaluating node importance is a critical aspect of analyzing complex systems, with broad applications in digital marketing, rumor suppression, and disease control. However, existing methods typically rely on conventional network structures and fail to capture the polyadic interactions intrinsic to many real-world systems. To address this limitation, we study key node identification in hypergraphs, where higher-order interactions are naturally modeled as hyperedges. We propose a novel framework, AHGA, which integrates an Autoencoder for extracting higher-order structural features, a HyperGraph neural network-based pre-training module (HGNN), and an Active learning-based fine-tuning process. This fine-tuning step plays a vital role in mitigating the gap between synthetic and real-world data, thereby enhancing the model's robustness and generalization across diverse hypergraph topologies. Extensive experiments on eight empirical hypergraphs show that AHGA outperforms classical centrality-based baselines by approximately 37.4%. Furthermore, the nodes identified by AHGA exhibit both high influence and strong structural disruption capability, demonstrating their superiority in detecting multifunctional nodes.

cs.SI↗

arXiv2025

Positive Style Accumulation: A Style Screening and Continuous Utilization Framework for Federated DG-ReID

The Federated Domain Generalization for Person re-identification (FedDG-ReID) aims to learn a global server model that can be effectively generalized to source and target domains through distributed source domain data. Existing methods mainly improve the diversity of samples through style transformation, which to some extent enhances the generalization performance of the model. However, we discover that not all styles contribute to the generalization performance. Therefore, we define styles that are beneficial or harmful to the model's generalization performance as positive or negative styles. Based on this, new issues arise: How to effectively screen and continuously utilize the positive styles. To solve these problems, we propose a Style Screening and Continuous Utilization (SSCU) framework. Firstly, we design a Generalization Gain-guided Dynamic Style Memory (GGDSM) for each client model to screen and accumulate generated positive styles. Meanwhile, we propose a style memory recognition loss to fully leverage the positive styles memorized by Memory. Furthermore, we propose a Collaborative Style Training (CST) strategy to make full use of positive styles. Unlike traditional learning strategies, our approach leverages both newly generated styles and the accumulated positive styles stored in memory to train client models on two distinct branches. This training strategy is designed to effectively promote the rapid acquisition of new styles by the client models, and guarantees the continuous and thorough utilization of positive styles, which is highly beneficial for the model's generalization performance. Extensive experimental results demonstrate that our method outperforms existing methods in both the source domain and the target domain.

cs.CV↗

arXiv2025

A Parameterized Barzilai-Borwein Method via Interpolated Least Squares

The Barzilai-Borwein (BB) method is an effective gradient descent algorithm for solving unconstrained optimization problems. Based on the observation of two classical BB step sizes, by constructing an interpolated least squares model, we propose a novel class of BB step sizes, each of which still retains the quasi-Newton property, with the original two BB step sizes being their two extreme cases. We present the mathematical principle underlying the adaptive alternating BB (ABB) method. Based on this principle, we develop a class of effective adaptive interpolation parameters. For strictly convex quadratic optimization problems, we establish the R-linear convergence of this new gradient descent method by investigating the evolution pattern of the ratio of the absolute values of the gradient components. Numerical experiments are conducted to illustrate our findings.

math.OC↗

arXiv2025

Electric-Field Reconstruction for Radio Detection of Inclined Air Showers in Three Polarizations

Accurate reconstruction of the electric field produced by extensive air showers is essential for the radio-detection technique, as the key parameters of interest of the primary particles that generated the showers are the amplitude, polarization, frequency spectrum, and energy fluence carried by the electric field at each receiving radio antenna. Conventional electric-field reconstruction methods primarily focus on antennas with two horizontal polarizations. In this work, we introduce an analytic $χ^2$ minimization method that is applicable to both two and three polarizations. This solution has been verified for simple and realistic antenna responses, with a particular focus on inclined air showers. Our method achieves standard deviations better than 4\% and 6\% for the estimation of the Hilbert peak envelope amplitude of the electric field and the energy fluence, respectively, with an antenna-response-dependent bias. Additionally, we have studied the dependence of the method with arrival direction showing that it has a good performance in the zenith range from 63$^\circ$ up to 80$^\circ$. This work also demonstrates that incorporating vertically polarized antennas enhances the precision of the reconstruction, leading to a more accurate and reliable electric-field estimation for inclined air showers.

astro-ph.IM↗

arXiv2025

Progress of the GRANDProto300 Project

GRANDProto300 (hereafter referred to as GP300) is a pioneering prototype array of the GRAND experiment. It consists of 300 radio antennas and will cover an area of 200 km$^2$ in a radio-quiet region of western China. Serving as a test bench for the GRAND experiment, GRANDProto300 aims to achieve autonomous radio detection and reconstruction of highly inclined air showers. It is designed to detect ultra-high-energy cosmic rays in the energy range of $10^{16.5}$-$10^{18}$ eV at a rate comparable to that of the Pierre Auger Observatory. Over the past two years, significant improvements have been made to both the hardware and firmware of GP300. Currently, 65 antenna units have been deployed at the site by June 2025. We present the current status of detector commissioning, including updates on hardware, calibration results such as GPS timing and antenna positioning. Additionally, we discuss the solar radio bursts associated with solar flares, the galactic radio emissions detected, and preliminary cosmic ray surveys.

astro-ph.HE↗

arXiv2025

Non-convergence of the principal eigenvalue of elliptic operators for large advection

This paper investigates the limit of the principal eigenvalue $λ(s)$ as $s\to+\infty$ for the following elliptic equation \begin{align*} -Δφ(x)-2s\mathbf{v}\cdot\nablaφ(x)+c(x)φ(x)=λ(s)φ(x), \quad x\in Ω \end{align*} in a bounded domain $Ω\subset \mathbb{R}^d (d\geq 1)$ with the Neumann boundary condition. Previous studies have shown that under certain conditions on $\mathbf{v}$, $λ(s)$ converges as $s\to\infty$ (including cases where $\lim\limits_{s \to\infty }λ(s)=\pm\infty$). This work constructs an example such that $λ(s)$ is divergent as $s\to+\infty$. This seems to be the first rigorous result demonstrating the non-convergence of the principal eigenvalue for second-order linear elliptic operators with some strong advection. As an application, we demonstrate that for the classical advection-reaction-diffusion model with advective velocity field $\mathbf{v}=\nabla m$, where $m$ is a potential function with infinite oscillations, the principal eigenvalue changes sign infinitely often along a subsequence of $s\to\infty$. This leads to solution behaviors that differ significantly from those observed when $m$ is non-oscillatory.

math.AP↗

arXiv2025

GPAS: Accelerating Convergence of LLM Pretraining via Gradient-Preserving Activation Scaling

Modern Large Language Models, such as the LLaMA, Qwen and DeepSeek series, predominantly adopt the Pre-LayerNorm (Pre-LN) Transformer architecture. While being stable during pretraining and scalable to large model sizes, Pre-LN suffers from an exponential growth in activation variance across layers, causing the shortcut to dominate over sub-layer outputs in the residual connection and limiting the learning capacity of deeper layers. To mitigate this issue, we propose Gradient-Preserving Activation Scaling (GPAS), a simple technique that can be used in combination with existing approaches. GPAS works by scaling down the intermediate activations while keeping their gradients unchanged. This leaves information in the activations intact, and avoids the gradient vanishing problem associated with gradient downscaling. Extensive experiments across various model sizes from 71M to 1B show that GPAS achieves consistent performance gains. Beyond enhancing Pre-LN Transformers, GPAS also shows promise in improving alternative architectures such as Sandwich-LN and DeepNorm, demonstrating its versatility and potential for improving training dynamics in a wide range of settings. Our code is available at https://github.com/dandingsky/GPAS.

cs.LG↗

arXiv2025

Few-shot Classification as Multi-instance Verification: Effective Backbone-agnostic Transfer across Domains

We investigate cross-domain few-shot learning under the constraint that fine-tuning of backbones (i.e., feature extractors) is impossible or infeasible -- a scenario that is increasingly common in practical use cases. Handling the low-quality and static embeddings produced by frozen, "black-box" backbones leads to a problem representation of few-shot classification as a series of multiple instance verification (MIV) tasks. Inspired by this representation, we introduce a novel approach to few-shot domain adaptation, named the "MIV-head", akin to a classification head that is agnostic to any pretrained backbone and computationally efficient. The core components designed for the MIV-head, when trained on few-shot data from a target domain, collectively yield strong performance on test data from that domain. Importantly, it does so without fine-tuning the backbone, and within the "meta-testing" phase. Experimenting under various settings and on an extension of the Meta-dataset benchmark for cross-domain few-shot image classification, using representative off-the-shelf convolutional neural network and vision transformer backbones pretrained on ImageNet1K, we show that the MIV-head achieves highly competitive accuracy when compared to state-of-the-art "adapter" (or partially fine-tuning) methods applied to the same backbones, while incurring substantially lower adaptation cost. We also find well-known "classification head" approaches lag far behind in terms of accuracy. Ablation study empirically justifies the core components of our approach. We share our code at https://github.com/xxweka/MIV-head.

cs.CV↗

arXiv2025

S^3cMath: Spontaneous Step-level Self-correction Makes Large Language Models Better Mathematical Reasoners

Self-correction is a novel method that can stimulate the potential reasoning abilities of large language models (LLMs). It involves detecting and correcting errors during the inference process when LLMs solve reasoning problems. However, recent works do not regard self-correction as a spontaneous and intrinsic capability of LLMs. Instead, such correction is achieved through post-hoc generation, external knowledge introduction, multi-model collaboration, and similar techniques. In this paper, we propose a series of mathematical LLMs called S$^3$c-Math, which are able to perform Spontaneous Step-level Self-correction for Mathematical reasoning. This capability helps LLMs to recognize whether their ongoing inference tends to contain errors and simultaneously correct these errors to produce a more reliable response. We proposed a method, which employs a step-level sampling approach to construct step-wise self-correction data for achieving such ability. Additionally, we implement a training strategy that uses above constructed data to equip LLMs with spontaneous step-level self-correction capacities. Our data and methods have been demonstrated to be effective across various foundation LLMs, consistently showing significant progress in evaluations on GSM8K, MATH, and other mathematical benchmarks. To the best of our knowledge, we are the first to introduce the spontaneous step-level self-correction ability of LLMs in mathematical reasoning.

cs.CL↗

arXiv2025

Foundation Model Insights and a Multi-Model Approach for Superior Fine-Grained One-shot Subset Selection

One-shot subset selection serves as an effective tool to reduce deep learning training costs by identifying an informative data subset based on the information extracted by an information extractor (IE). Traditional IEs, typically pre-trained on the target dataset, are inherently dataset-dependent. Foundation models (FMs) offer a promising alternative, potentially mitigating this limitation. This work investigates two key questions: (1) Can FM-based subset selection outperform traditional IE-based methods across diverse datasets? (2) Do all FMs perform equally well as IEs for subset selection? Extensive experiments uncovered surprising insights: FMs consistently outperform traditional IEs on fine-grained datasets, whereas their advantage diminishes on coarse-grained datasets with noisy labels. Motivated by these finding, we propose RAM-APL (RAnking Mean-Accuracy of Pseudo-class Labels), a method tailored for fine-grained image datasets. RAM-APL leverages multiple FMs to enhance subset selection by exploiting their complementary strengths. Our approach achieves state-of-the-art performance on fine-grained datasets, including Oxford-IIIT Pet, Food-101, and Caltech-UCSD Birds-200-2011.

cs.CV↗

arXiv2025

LAECIPS: Large Vision Model Assisted Adaptive Edge-Cloud Collaboration for IoT-based Embodied Intelligence System

Embodied intelligence (EI) enables manufacturing systems to flexibly perceive, reason, adapt, and operate within dynamic shop floor environments. In smart manufacturing, a representative EI scenario is robotic visual inspection, where industrial robots must accurately inspect components on rapidly changing, heterogeneous production lines. This task requires both high inference accuracy especially for uncommon defects and low latency to match production speeds, despite evolving lighting, part geometries, and surface conditions. To meet these needs, we propose LAECIPS, a large vision model-assisted adaptive edge-cloud collaboration framework for IoT-based embodied intelligence systems. LAECIPS decouples large vision models in the cloud from lightweight models on the edge, enabling plug-and-play model adaptation and continual learning. Through a hard input mining-based inference strategy, LAECIPS routes complex and uncertain inspection cases to the cloud while handling routine tasks at the edge, achieving both high accuracy and low latency. Experiments conducted on a real-world robotic semantic segmentation system for visual inspection demonstrate significant improvements in accuracy, processing latency, and communication overhead compared to state-of-the-art methods. LAECIPS provides a practical and scalable foundation for embodied intelligence in smart manufacturing, especially in adaptive robotic inspection and quality control scenarios.

cs.AI↗

arXiv2025

StereoVAE: A lightweight stereo-matching system using embedded GPUs

We present a lightweight system for stereo matching through embedded GPUs. It breaks the trade-off between accuracy and processing speed in stereo matching, enabling our embedded system to further improve the matching accuracy while ensuring real-time processing. The main idea of our method is to construct a tiny neural network based on variational auto-encoder (VAE) to upsample and refinement a small size of coarse disparity map, which is first generated by a traditional matching method. The proposed hybrid structure cannot only bring the advantage of traditional methods in terms of computational complexity, but also ensure the matching accuracy under the impact of neural network. Extensive experiments on the KITTI 2015 benchmark demonstrate that our tiny system exhibits high robustness in improving the accuracy of the coarse disparity maps generated by different algorithms, while also running in real-time on embedded GPUs.

cs.CV↗

arXiv2025

Safe: Enhancing Mathematical Reasoning in Large Language Models via Retrospective Step-aware Formal Verification

Chain-of-Thought (CoT) prompting has become the de facto method to elicit reasoning capabilities from large language models (LLMs). However, to mitigate hallucinations in CoT that are notoriously difficult to detect, current methods such as process reward models (PRMs) or self-consistency operate as opaque boxes and do not provide checkable evidence for their judgments, possibly limiting their effectiveness. To address this issue, we draw inspiration from the idea that "the gold standard for supporting a mathematical claim is to provide a proof". We propose a retrospective, step-aware formal verification framework $Safe$. Rather than assigning arbitrary scores, we strive to articulate mathematical claims in formal mathematical language Lean 4 at each reasoning step and provide formal proofs to identify hallucinations. We evaluate our framework $Safe$ across multiple language models and various mathematical datasets, demonstrating a significant performance improvement while offering interpretable and verifiable evidence. We also propose $FormalStep$ as a benchmark for step correctness theorem proving with $30,809$ formal statements. To the best of our knowledge, our work represents the first endeavor to utilize formal mathematical language Lean 4 for verifying natural language content generated by LLMs, aligning with the reason why formal mathematical languages were created in the first place: to provide a robust foundation for hallucination-prone human-written proofs.

cs.CL↗