检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Xu, Shu Cao, Liqun School of Mathematical Sciences Peking University Beijing100871 China Institute of Computational Mathematics and Scientific/Engineering Computing Academy of Mathematics and Systems Science Chinese Academy of Sciences Beijing100190 China State Key Laboratory of Scientific and Engineering Computing Beijing100190 China National Center for Mathematics and Interdisciplinary Sciences Chinese Academy of Sciences Beijing100190 China

We investigate the convergence of a backward Euler finite element discretization applied to a multi-domain and multi-scale elliptic-parabolic system, derived from the Doyle-Fuller-Newman model for lithium-ion batteries. We establish optimal-order error estimates for the solution in the norms l2(H1) and l2(L2(Hrq)), q = 0,1. To improve computational efficiency, we propose a novel solver that accelerates the solution process and controls memory usage. Numerical experiments with realistic battery parameters validate the theoretical error rates and demonstrate the significantly superior performance of the proposed solver over existing solvers. Copyright © 2024, The Authors. All rights reserved.

关键词： Computational efficiency

Visual Class Incremental Learning with Textual Priors Guidance based on an Adapted Vision-Language Model

学校读者我要写书评

暂无评论

IEEE Transactions on Multimedia 2025年

作者： Zhang, Wentao Yu, Tong Wang, Ruixuan Xie, Jianhui Trucco, Emanuele Zheng, Wei-Shi Yang, Xiaobo Sun Yat-sen Univerisity School of Computer Science and Engineering Guangzhou510275 China Peng Cheng Laboratory Shenzhen518066 China MOE Key Laboratory of Machine Intelligence and Advanced Computing Guangzhou510275 China Second Affiliated Hospital Guangzhou University of Chinese Medicine State Key Laboratory of Dampness Syndrome of Chinese Medicine Guangzhou510260 China Guangdong Provincial Key Laboratory of Clinical Research on Traditional Chinese Medicine Syndrome Guangzhou510120 China University of Dundee School of Science and Engineering DundeeDD1 4HN United Kingdom

An ideal artificial intelligence (AI) system should have the capability to continually learn like humans. However, when learning new knowledge, AI systems often suffer from catastrophic forgetting of old knowledge. Although many continual learning methods have been proposed, they often ignore the issue of misclassifying similar classes and make insufficient use of textual priors of visual classes to improve continual learning performance. In this study, we propose a continual learning framework based on a pre-trained vision-language model (VLM) that does not require storing old class data. This framework utilizes parameter-efficient fine-tuning of the VLM's text encoder for constructing a shared and consistent semantic textual space throughout the continual learning process. The textual priors of visual classes are encoded by the adapted VLM's text encoder to generate discriminative semantic representations, which are then used to guide the learning of visual classes. Additionally, fake out-of-distribution (OOD) images constructed from each training image further assist in the learning of visual classes. Extensive empirical evaluations on three natural datasets and one medical dataset demonstrate the superiority of the proposed framework. The source code is available at https://***/OpenMedIA/CIL_Adapterd_VLM. © 1999-2012 IEEE.

关键词： Contrastive Learning

ARIW-Framework: Adaptive Robust Iterative Watermarking Framework

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Wu, Shaowu Lu, Wei Zeng, Liting Luo, Xiangyang School of Computer Science and Engineering Sun Yat-sen University Guangzhou China State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

With the rapid rise of large models, copyright protection for generated image content has become a critical security challenge. Although deep learning watermarking techniques offer an effective solution for digital image copyright protection, they still face limitations in terms of visual quality, robustness and generalization. To address these issues, this paper proposes an adaptive robust iterative watermarking framework (ARIW-Framework) that achieves high-quality watermarked images while maintaining exceptional robustness and generalization performance. Specifically, we introduce an iterative approach to optimize the encoder for generating robust residuals. The encoder incorporates noise layers and a decoder to compute robustness weights for residuals under various noise attacks. By employing a parallel optimization strategy, the framework enhances robustness against multiple types of noise attacks. Furthermore, we leverage image gradients to determine the embedding strength at each pixel location, significantly improving the visual quality of the watermarked images. Extensive experiments demonstrate that the proposed method achieves superior visual quality while exhibiting remarkable robustness and generalization against noise attacks. Copyright © 2025, The Authors. All rights reserved.

关键词： Image watermarking

Diffusion-based Adversarial Identity Manipulation for Facial Privacy Protection

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Wang, Liqin Lu, Wei Hu, Qianyue Luo, Xiangyang School of Computer Science and Engineering Sun Yat-sen University Guangzhou China State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

The success of face recognition (FR) systems has led to serious privacy concerns due to potential unauthorized surveillance and user tracking on social networks. Existing methods for enhancing privacy fail to generate natural face images that can protect facial privacy. In this paper, we propose diffusion-based adversarial identity manipulation (DiffAIM) to generate natural and highly transferable adversarial faces against malicious FR systems. To be specific, we manipulate facial identity within the low-dimensional latent space of a diffusion model. This involves iteratively injecting gradient-based adversarial identity guidance during the reverse diffusion process, progressively steering the generation toward the desired adversarial faces. The guidance is optimized for identity convergence towards a target while promoting semantic divergence from the source, facilitating effective impersonation while maintaining visual naturalness. We further incorporate structure-preserving regularization to preserve facial structure consistency during manipulation. Extensive experiments on both face verification and identification tasks demonstrate that compared with the state-of-the-art, DiffAIM achieves stronger black-box attack transferability while maintaining superior visual quality. We also demonstrate the effectiveness of the proposed approach for commercial FR APIs, including Face++ and Aliyun. Copyright © 2025, The Authors. All rights reserved.

关键词： Computer forensics

A Security Sensitive Function Mining Approach Based on Precondition Pattern Analysis

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2020年第5期63卷 1013-1029页

作者： Zhongxu Yin Yiran Song Huiqin Chen Yan Cao State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou450001China Henan University of Animal Husbandry&Economy Zhengzhou450046China University of Michigan Transportation Research Institute Michigan48109-2150USA Zhengzhou University Zhengzhou450001China

Security-sensitive functions are the basis for building a taint-style vulnerability *** approaches for extracting security-sensitive functions either don’t analyze data flow accurately,or not conducting pattern analyzing of conditions,resulting in higher false positive rate or false negative rate,which increased manual confirmation *** this paper,we propose a security sensitive function mining approach based on preconditon pattern ***,we propose an enhanced system dependency graph analysis algorithm for precisely extracting the conditional statements which check the function parameters and conducting statistical analysis of the conditional statements for selecting candidate security sensitive functions of the target *** we adopt a precondition pattern mining method based on conditional statements nomalizing and *** with fixed precondition patterns are regarded as security-sensitive *** experimental results on four popular open source codebases of different scales show that the approach proposed is effective in reducing the false positive rate and false negative rate for detecting security sensitive functions.

关键词： Code mining security sensitive function function preconditions single-linkage clustering

A Novel Method for Detection of ECG with Deep Learning 7

学校读者我要写书评

暂无评论

A Novel Method for Detection of ECG with Deep Learning

7th International Conference on Computer and Communications, ICCC 2021

作者： Wang, Haiyan Zhou, Yanjie Zhou, Bing Wang, Zongmin State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China Simulation Experiment Centre Zhengzhou University of Aeronautics Zhengzhou China Collaborative Innovation Centre For Internet Healthcare Zhengzhou University Zhengzhou China School of Management Engineering Zhengzhou University Zhengzhou China School of Information Engineering Zhengzhou University Zhengzhou China

ISBN: (纸本)9781665409506

With the rapid development of machine learning and deep learning, ECG intelligent detection models lean more heavily on the labeled data. However, the development of ECG annotation cannot satisfy the development of an intelligent detection model because of its high complexity and heavy workload. Developing a model with low dependence on labeled data becomes the core for solving the ECG intelligent detection. To overcome this limitation, this paper proposes a deep active learning (DAL) model for ECG intelligent detection, which combines high-performance deep learning (DL) with active learning (AL) that is less dependent on labeled data. Experimental results show that the proposed method can achieve high performance with the support of a small amount of labeled data. The proposed method could be adopted for data annotation, which achieves a high-quality data annotation and avoids useless annotation. © 2021 IEEE.

关键词： Electrocardiograms

Real-Time Source-Independent Quantum Random Number Generator Based on a Cloud Superconducting Quantum Computer

学校读者我要写书评

暂无评论

Research Square

Research Square 2021年

作者： Li, Yuanhao Fei, Yangyang Wang, Weilong Meng, Xiangdong Wang, Hong Duan, Qianheng Ma, Zhi State Key Laboratory of Mathematical Engineering and Advanced Computing Henan Zhengzhou450001 China Henan Key Laboratory of Network Cryptography Technology Henan Zhengzhou450001 China

Quantum random number generator (QRNG) relies on the intrinsic randomness of quantum mechanics to produce true random numbers which are important in information processing tasks. Due to the presence of the superposition state, quantum computer can be used as a true random number generator. However, in practice, the implementation of quantum computer is subject to various noise sources which affect the randomness of the generated random numbers. To solve this problem, we propose a source-independent QRNG (SI-QRNG) scheme based on quantum computer which is motivated by the SI-QRNG scheme in quantum optics. The scheme can provide certified randomness by estimating the preparation error of superposition states in real time even when the source is untrusted, under the assumption that the measurement operation is trusted. Our analysis takes into account the readout error of quantum state and further gives the final extracted number of random bits. And the estimation method of preparation error of superposition state in randomness source. We also provide a parameter optimization method to increase the generation rate of random bits. In addition, by utilizing the cloud superconducting quantum computer of IBM, we experimentally demonstrate the practicality of our SI-QRNG scheme and achieve the generation of true random numbers. © 2021, CC BY.

关键词： Errors

A RIEMANNIAN EXPONENTIAL AUGMENTED LAGRANGIAN METHOD FOR computing THE PROJECTION ROBUST WASSERSTEIN DISTANCE

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Jiang, Bo Liu, Ya-Feng Key Laboratory for NSLSCS of Jiangsu Province School of Mathematical Sciences Nanjing Normal University Nanjing China State Key Laboratory of Scientific and Engineering Computing Institute of Computational Mathematics and Scientific/Engineering Computing Academy of Mathematics and Systems Science Chinese Academy of Sciences Beijing China

Projecting the distance measures onto a low-dimensional space is an efficient way of mitigating the curse of dimensionality in the classical Wasserstein distance using optimal transport. The obtained maximized distance is referred to as projection robust Wasserstein (PRW) distance. In this paper, we equivalently reformulate the computation of the PRW distance as an optimization problem over the Cartesian product of the Stiefel manifold and the Euclidean space with additional nonlinear inequality constraints. We propose a Riemannian exponential augmented Lagrangian method (ReALM) with a global convergence guarantee to solve this problem. Compared with the existing approaches, ReALM can potentially avoid too small penalty parameters. Moreover, we propose a framework of inexact Riemannian gradient descent methods to solve the subproblems in ReALM efficiently. In particular, by using the special structure of the subproblem, we give a practical algorithm named as the inexact Riemannian Barzilai-Borwein method with Sinkhorn iteration (iRBBS). The remarkable features of iRBBS lie in that it performs a flexible number of Sinkhorn iterations to compute an inexact gradient with respect to the projection matrix of the problem and adopts the Barzilai-Borwein stepsize based on the inexact gradient information to improve the performance. We show that iRBBS can return an ǫ-stationary point of the original PRW distance problem within O(ǫ−3) iterations. Extensive numerical results on synthetic and real datasets demonstrate that our proposed ReALM as well as iRBBS outperform the state-of-the-art solvers for computing the PRW *** Codes 65K10, 90C26, 90C47 Copyright © 2022, The Authors. All rights reserved.

关键词： Constrained optimization

LEFV: A Lightweight and Efficient System for Face Verification with Deep Convolution Neural Networks 19

学校读者我要写书评

暂无评论

LEFV: A Lightweight and Efficient System for Face Verificati...

Proceedings of the 3rd International Conference on Video and Image Processing

作者： Ming Liu Ping Zhang Qingbao Li Jinjin Liu Zhifeng Chen State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

ISBN: (纸本)9781450376822

The emergence of deep learning has made great progress in face recognition. With the popularization of embedded devices, deploying the deep model on embedded devices has become a trend. Most high-precision models require lots of computation costs. Therefore, developing a lightweight deep face recognition system running on embedded devices is a hot topic in current research. To achieve high-accuracy real-time performance of an embedded device, we now present a simple and effective face recognition system LEFV, including face detection, face normalization and face recognition. The quantitative experiments on two large-scale challenging datasets, WIDER FACE dataset and IJB-A dataset, show competitive performances on both runtime and accuracy.

关键词： Face Verification Face Detection Deep Convolution Neural Networks Lightweight Model

Progressive Multi-Stage Feature Mix for Person Re-Identification

学校读者我要写书评

暂无评论

Progressive Multi-Stage Feature Mix for Person Re-Identifica...

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Yan Zhang Binyu He Li Sun Qingli Li Shanghai Key Laboratory of Multidimensional Information Processing East China Normal University Shanghai China State Key Laboratory of Mathematical Engineering and Advanced Computing Wuxi China

Image features from a small local region often give strong evidence in person re-identification task. However, CNN suffers from paying too much attention on the most salient local areas, thus ignoring other discriminative clues, e.g., hair, shoes or logos on clothes. In this work, we propose a Progressive Multi-stage feature Mix network (PMM), which enables the model to find out the more precise and diverse features in a progressive manner. Specifically, (i) to enforce the model to look for different clues in the image, we adopt a multi-stage classifier and expect that the model is able to focus on a complementary region in each stage. (ii) we propose an Attentive feature Hard-Mix (A-Hard-Mix) to replace the salient feature blocks by the negative example in the current batch, whose label is different from the current sample. (iii) extensive experiments have been carried out on reID datasets such as the Market-1501, DukeMTMC-reID and CUHK03, showing that the proposed method can boost the re-identification performance significantly. Source code¹ has been released.

关键词： Training Hair Conferences Footwear Signal processing Acoustics Task analysis