Search arXiv⌕ Search

arXiv subjects

Hui Xue

Publications and source records attributed to Hui Xue.

At least 91 records · Page 5Linked to original sources
arXiv2025

Non-repetition of second coefficients of Hecke polynomials

Let $T_m(N,2k)$ denote the $m$-th Hecke operator on the space $S_{2k}(Γ_0(N))$ of cuspidal modular forms of weight $2k$ and level $N$. In this paper, we study the non-repetition of the second coefficient of the characteristic polynomial of $T_m(N,2k)$. We obtain results in the horizontal aspect (where $m$ varies), the vertical aspect (where $k$ varies), and the level aspect (where $N$ varies). Finally, we use these non-repetition results to extend a result of Vilardi and Xue on distinguishing Hecke eigenforms.

math.NT↗
arXiv2025

Score-based Generative Modeling for Conditional Independence Testing

Determining conditional independence (CI) relationships between random variables is a fundamental yet challenging task in machine learning and statistics, especially in high-dimensional settings. Existing generative model-based CI testing methods, such as those utilizing generative adversarial networks (GANs), often struggle with undesirable modeling of conditional distributions and training instability, resulting in subpar performance. To address these issues, we propose a novel CI testing method via score-based generative modeling, which achieves precise Type I error control and strong testing power. Concretely, we first employ a sliced conditional score matching scheme to accurately estimate conditional score and use Langevin dynamics conditional sampling to generate null hypothesis samples, ensuring precise Type I error control. Then, we incorporate a goodness-of-fit stage into the method to verify generated samples and enhance interpretability in practice. We theoretically establish the error bound of conditional distributions modeled by score-based generative models and prove the validity of our CI tests. Extensive experiments on both synthetic and real-world datasets show that our method significantly outperforms existing state-of-the-art methods, providing a promising way to revitalize generative model-based CI testing.

cs.LG↗
arXiv2025

Towards the Resistance of Neural Network Watermarking to Fine-tuning

This paper proves a new watermarking method to embed the ownership information into a deep neural network (DNN), which is robust to fine-tuning. Specifically, we prove that when the input feature of a convolutional layer only contains low-frequency components, specific frequency components of the convolutional filter will not be changed by gradient descent during the fine-tuning process, where we propose a revised Fourier transform to extract frequency components from the convolutional filter. Additionally, we also prove that these frequency components are equivariant to weight scaling and weight permutations. In this way, we design a watermark module to encode the watermark information to specific frequency components in a convolutional filter. Preliminary experiments demonstrate the effectiveness of our method.

cs.LG↗
arXiv2025

Imaging Transformer for MRI Denoising: a Scalable Model Architecture that enables SNR << 1 Imaging

Purpose: To propose a flexible and scalable imaging transformer (IT) architecture with three attention modules for multi-dimensional imaging data and apply it to MRI denoising with very low input SNR. Methods: Three independent attention modules were developed: spatial local, spatial global, and frame attentions. They capture long-range signal correlation and bring back the locality of information in images. An attention-cell-block design processes 5D tensors ([B, C, F, H, W]) for 2D, 2D+T, and 3D image data. A High Resolution (HRNet) backbone was built to hold IT blocks. Training dataset consists of 206,677 cine series and test datasets had 7,267 series. Ten input SNR levels from 0.05 to 8.0 were tested. IT models were compared to seven convolutional and transformer baselines. To test scalability, four IT models 27m to 218m parameters were trained. Two senior cardiologists reviewed IT model outputs from which the EF was measured and compared against the ground-truth. Results: IT models significantly outperformed other models over the tested SNR levels. The performance gap was most prominent at low SNR levels. The IT-218m model had the highest SSIM and PSNR, restoring good image quality and anatomical details even at SNR 0.2. Two experts agreed at this SNR or above, the IT model output gave the same clinical interpretation as the ground-truth. The model produced images that had accurate EF measurements compared to ground-truth values. Conclusions: Imaging transformer model offers strong performance, scalability, and versatility for MR denoising. It recovers image quality suitable for confident clinical reading and accurate EF measurement, even at very low input SNR of 0.2.

eess.IV↗
arXiv2025

The Right Time Matters: Data Arrangement Affects Zero-Shot Generalization in Instruction Tuning

Understanding alignment techniques begins with comprehending zero-shot generalization brought by instruction tuning, but little of the mechanism has been understood. Existing work has largely been confined to the task level, without considering that tasks are artificially defined and, to LLMs, merely consist of tokens and representations. To bridge this gap, we investigate zero-shot generalization from the perspective of the data itself. We first demonstrate that zero-shot generalization happens very early during instruction tuning, with loss serving as a stable indicator. Next, we investigate training data arrangement through similarity and granularity perspectives, confirming that the timing of exposure to certain training examples may greatly facilitate generalization on unseen tasks. Finally, we propose a more grounded training data arrangement framework, Test-centric Multi-turn Arrangement, and show its effectiveness in promoting continual learning and further loss reduction. For the first time, we show that zero-shot generalization during instruction tuning is a form of similarity-based generalization between training and test data at the instance level. Our code is released at https://github.com/thunlp/Dynamics-of-Zero-Shot-Generalization.

cs.CL↗
arXiv2025

S-Eval: Towards Automated and Comprehensive Safety Evaluation for Large Language Models

Generative large language models (LLMs) have revolutionized natural language processing with their transformative and emergent capabilities. However, recent evidence indicates that LLMs can produce harmful content that violates social norms, raising significant concerns regarding the safety and ethical ramifications of deploying these advanced models. Thus, it is both critical and imperative to perform a rigorous and comprehensive safety evaluation of LLMs before deployment. Despite this need, owing to the extensiveness of LLM generation space, it still lacks a unified and standardized risk taxonomy to systematically reflect the LLM content safety, as well as automated safety assessment techniques to explore the potential risk efficiently. To bridge the striking gap, we propose S-Eval, a novel LLM-based automated Safety Evaluation framework with a newly defined comprehensive risk taxonomy. S-Eval incorporates two key components, i.e., an expert testing LLM ${M}_t$ and a novel safety critique LLM ${M}_c$. ${M}_t$ is responsible for automatically generating test cases in accordance with the proposed risk taxonomy. ${M}_c$ can provide quantitative and explainable safety evaluations for better risk awareness of LLMs. In contrast to prior works, S-Eval is efficient and effective in test generation and safety evaluation. Moreover, S-Eval can be flexibly configured and adapted to the rapid evolution of LLMs and accompanying new safety threats, test generation methods and safety critique methods thanks to the LLM-based architecture. S-Eval has been deployed in our industrial partner for the automated safety evaluation of multiple LLMs serving millions of users, demonstrating its effectiveness in real-world scenarios. Our benchmark is publicly available at https://github.com/IS2Lab/S-Eval.

cs.CR↗
arXiv2025

SNRAware: Improved Deep Learning MRI Denoising with SNR Unit Training and G-factor Map Augmentation

To develop and evaluate a new deep learning MR denoising method that leverages quantitative noise distribution information from the reconstruction process to improve denoising performance and generalization. This retrospective study trained 14 different transformer and convolutional models with two backbone architectures on a large dataset of 2,885,236 images from 96,605 cardiac retro-gated cine complex series acquired at 3T. The proposed training scheme, termed SNRAware, leverages knowledge of the MRI reconstruction process to improve denoising performance by simulating large, high quality, and diverse synthetic datasets, and providing quantitative information about the noise distribution to the model. In-distribution testing was performed on a hold-out dataset of 3000 samples with performance measured using PSNR and SSIM, with ablation comparison without the noise augmentation. Out-of-distribution tests were conducted on cardiac real-time cine, first-pass cardiac perfusion, and neuro and spine MRI, all acquired at 1.5T, to test model generalization across imaging sequences, dynamically changing contrast, different anatomies, and field strengths. The best model found in the in-distribution test generalized well to out-of-distribution samples, delivering 6.5x and 2.9x CNR improvement for real-time cine and perfusion imaging, respectively. Further, a model trained with 100% cardiac cine data generalized well to a T1 MPRAGE neuro 3D scan and T2 TSE spine MRI.

physics.med-ph↗
arXiv2025

EasyRobust: A Comprehensive and Easy-to-use Toolkit for Robust and Generalized Vision

Deep neural networks (DNNs) has shown great promise in computer vision tasks. However, machine vision achieved by DNNs cannot be as robust as human perception. Adversarial attacks and data distribution shifts have been known as two major scenarios which degrade machine performance and obstacle the wide deployment of machines "in the wild". In order to break these obstructions and facilitate the research of model robustness, we develop EasyRobust, a comprehensive and easy-to-use toolkit for training, evaluation and analysis of robust vision models. EasyRobust targets at two types of robustness: 1) Adversarial robustness enables the model to defense against malicious inputs crafted by worst-case perturbations, also known as adversarial examples; 2) Non-adversarial robustness enhances the model performance on natural test images with corruptions or distribution shifts. Thorough benchmarks on image classification enable EasyRobust to provide an accurate robustness evaluation on vision models. We wish our EasyRobust can help for training practically-robust models and promote academic and industrial progress in closing the gap between human and machine vision. Codes and models of EasyRobust have been open-sourced in https://github.com/alibaba/easyrobust.

cs.CV↗
arXiv2025

Lifelong Knowledge Editing for LLMs with Retrieval-Augmented Continuous Prompt Learning

Model editing aims to correct outdated or erroneous knowledge in large language models (LLMs) without the need for costly retraining. Lifelong model editing is the most challenging task that caters to the continuous editing requirements of LLMs. Prior works primarily focus on single or batch editing; nevertheless, these methods fall short in lifelong editing scenarios due to catastrophic knowledge forgetting and the degradation of model performance. Although retrieval-based methods alleviate these issues, they are impeded by slow and cumbersome processes of integrating the retrieved knowledge into the model. In this work, we introduce RECIPE, a RetriEval-augmented ContInuous Prompt lEarning method, to boost editing efficacy and inference efficiency in lifelong learning. RECIPE first converts knowledge statements into short and informative continuous prompts, prefixed to the LLM's input query embedding, to efficiently refine the response grounded on the knowledge. It further integrates the Knowledge Sentinel (KS) that acts as an intermediary to calculate a dynamic threshold, determining whether the retrieval repository contains relevant knowledge. Our retriever and prompt encoder are jointly trained to achieve editing properties, i.e., reliability, generality, and locality. In our experiments, RECIPE is assessed extensively across multiple LLMs and editing datasets, where it achieves superior editing performance. RECIPE also demonstrates its capability to maintain the overall performance of LLMs alongside showcasing fast editing and inference speed.

cs.CL↗
arXiv2025

Asymptotics and sign patterns of Hecke polynomial coefficients

We determine the asymptotic behavior of the coefficients of Hecke polynomials. In particular, this allows us to determine signs of these coefficients when the level or the weight is sufficiently large. In all but finitely many cases, this also verifies a conjecture on the nanvanishing of the coefficients of Hecke polynomials.

math.NT↗
arXiv2025

Nonvanishing of Second Coefficients of Hecke Polynomials on the Newspace

For $m \geq 1$, let $N \geq 1$ be coprime to $m$, $k \geq 2$, and $χ$ be a Dirichlet character modulo $N$ with $χ(-1)=(-1)^k$. Then let $T_m^{\text{new}}(N,k,χ)$ denote the restriction of the $m$-th Hecke operator to the space $S_k^{\text{new}}(Γ_0(N), χ)$. We demonstrate that for fixed $m$ and trivial character $χ$, the second coefficient of the characteristic polynomial of $T_m^{\text{new}}(N,k)$ vanishes for only finitely many pairs $(N,k)$, and we further determine the sign. To demonstrate our method, for $m=2,4$, we also compute all pairs $(N,k)$ for which the second coefficient vanishes. In the general character case, we also show that excluding an infinite family where $S_k^{\text{new}}(Γ_0(N), χ)$ is trivial, the second coefficient of the characteristic polynomial of $T_m^{\text{new}}(N,k,χ)$ vanishes for only finitely many triples $(N,k,χ)$.

math.NT↗
arXiv2025

QExplorer: Large Language Model Based Query Extraction for Toxic Content Exploration

Automatically extracting effective queries is challenging in information retrieval, especially in toxic content exploration, as such content is likely to be disguised. With the recent achievements in generative Large Language Model (LLM), we are able to leverage the capabilities of LLMs to extract effective queries for similar content exploration directly. This study proposes QExplorer, an approach of large language model based Query Extraction for toxic content Exploration. The QExplorer approach involves a 2-stage training process: instruction Supervised FineTuning (SFT) and preference alignment using Direct Preference Optimization (DPO), as well as the datasets construction with feedback of search system. To verify the effectiveness of QExplorer, a series of offline and online experiments are conducted on our real-world system. The offline empirical results demonstrate that the performance of our automatic query extraction outperforms that of several LLMs and humans. The online deployment shows a significant increase in the detection of toxic items.

cs.IR↗
arXiv2025

Zeros of symmetric power period polynomials

Suppose that $k$ and $N$ are positive integers. Let $f$ be a newform on $Γ_0(N)$ of weight $k$ with $L$-function $L_f(s)$. Previous works have studied the zeros of the period polynomial $r_f(z)$, which is a generating function for the critical values of $L_f(s)$ and has a functional equation relating $z$ and $-1/Nz$. In particular, $r_f(z)$ satisfies a version of the Riemann hypothesis: all of its zeros are on the circle of symmetry $\{z \in \C \ : \ |z|=1/\sqrt{N}\}$. In this paper, for a positive integer $m$, we define a natural analogue of $r_f(z)$ for the $m^{\operatorname{th}}$ symmetric power $L$-function of $f$ when $N$ is squarefree. Our analogue also has a functional equation relating $z$ and $-1/Nz$. We prove the corresponding version of the Riemann hypothesis when $k$ is large enough. Moreover, when $k>2(\operatorname{log}_2(13e^{2π}/9)+m)+1$, we prove our result when $N$ is large enough.

math.NT↗
arXiv2025

Mirage in the Eyes: Hallucination Attack on Multi-modal Large Language Models with Only Attention Sink

Fusing visual understanding into language generation, Multi-modal Large Language Models (MLLMs) are revolutionizing visual-language applications. Yet, these models are often plagued by the hallucination problem, which involves generating inaccurate objects, attributes, and relationships that do not match the visual content. In this work, we delve into the internal attention mechanisms of MLLMs to reveal the underlying causes of hallucination, exposing the inherent vulnerabilities in the instruction-tuning process. We propose a novel hallucination attack against MLLMs that exploits attention sink behaviors to trigger hallucinated content with minimal image-text relevance, posing a significant threat to critical downstream applications. Distinguished from previous adversarial methods that rely on fixed patterns, our approach generates dynamic, effective, and highly transferable visual adversarial inputs, without sacrificing the quality of model responses. Comprehensive experiments on 6 prominent MLLMs demonstrate the efficacy of our attack in compromising black-box MLLMs even with extensive mitigating mechanisms, as well as the promising results against cutting-edge commercial APIs, such as GPT-4o and Gemini 1.5. Our code is available at https://huggingface.co/RachelHGF/Mirage-in-the-Eyes.

cs.LG↗
arXiv2025

MRJ-Agent: An Effective Jailbreak Agent for Multi-Round Dialogue

Large Language Models (LLMs) demonstrate outstanding performance in their reservoir of knowledge and understanding capabilities, but they have also been shown to be prone to illegal or unethical reactions when subjected to jailbreak attacks. To ensure their responsible deployment in critical applications, it is crucial to understand the safety capabilities and vulnerabilities of LLMs. Previous works mainly focus on jailbreak in single-round dialogue, overlooking the potential jailbreak risks in multi-round dialogues, which are a vital way humans interact with and extract information from LLMs. Some studies have increasingly concentrated on the risks associated with jailbreak in multi-round dialogues. These efforts typically involve the use of manually crafted templates or prompt engineering techniques. However, due to the inherent complexity of multi-round dialogues, their jailbreak performance is limited. To solve this problem, we propose a novel multi-round dialogue jailbreaking agent, emphasizing the importance of stealthiness in identifying and mitigating potential threats to human values posed by LLMs. We propose a risk decomposition strategy that distributes risks across multiple rounds of queries and utilizes psychological strategies to enhance attack strength. Extensive experiments show that our proposed method surpasses other attack methods and achieves state-of-the-art attack success rate. We will make the corresponding code and dataset available for future research. The code will be released soon.

cs.AI↗
arXiv2025

KAnoCLIP: Zero-Shot Anomaly Detection through Knowledge-Driven Prompt Learning and Enhanced Cross-Modal Integration

Zero-shot anomaly detection (ZSAD) identifies anomalies without needing training samples from the target dataset, essential for scenarios with privacy concerns or limited data. Vision-language models like CLIP show potential in ZSAD but have limitations: relying on manually crafted fixed textual descriptions or anomaly prompts is time-consuming and prone to semantic ambiguity, and CLIP struggles with pixel-level anomaly segmentation, focusing more on global semantics than local details. To address these limitations, We introduce KAnoCLIP, a novel ZSAD framework that leverages vision-language models. KAnoCLIP combines general knowledge from a Large Language Model (GPT-3.5) and fine-grained, image-specific knowledge from a Visual Question Answering system (Llama3) via Knowledge-Driven Prompt Learning (KnPL). KnPL uses a knowledge-driven (KD) loss function to create learnable anomaly prompts, removing the need for fixed text prompts and enhancing generalization. KAnoCLIP includes the CLIP visual encoder with V-V attention (CLIP-VV), Bi-Directional Cross-Attention for Multi-Level Cross-Modal Interaction (Bi-CMCI), and Conv-Adapter. These components preserve local visual semantics, improve local cross-modal fusion, and align global visual features with textual information, enhancing pixel-level anomaly detection. KAnoCLIP achieves state-of-the-art performance in ZSAD across 12 industrial and medical datasets, demonstrating superior generalization compared to existing methods.

cs.CV↗
arXiv2024

A Study on Context Length and Efficient Transformers for Biomedical Image Analysis

Biomedical imaging modalities often produce high-resolution, multi-dimensional images that pose computational challenges for deep neural networks. These computational challenges are compounded when training transformers due to the self-attention operator, which scales quadratically with context length. Recent developments in long-context models have potential to alleviate these difficulties and enable more efficient application of transformers to large biomedical images, although a systematic evaluation on this topic is lacking. In this study, we investigate the impact of context length on biomedical image analysis and we evaluate the performance of recently proposed long-context models. We first curate a suite of biomedical imaging datasets, including 2D and 3D data for segmentation, denoising, and classification tasks. We then analyze the impact of context length on network performance using the Vision Transformer and Swin Transformer by varying patch size and attention window size. Our findings reveal a strong relationship between context length and performance, particularly for pixel-level prediction tasks. Finally, we show that recent long-context models demonstrate significant improvements in efficiency while maintaining comparable performance, though we highlight where gaps remain. This work underscores the potential and challenges of using long-context models in biomedical imaging.

cs.CV↗