检索结果-内蒙古大学图书馆

arXiv 2022年

作者： Jin, Diancong Chen, Ziheng Zhou, Tau School of Mathematics and Statistics Huazhong University of Science and Technology Wuhan430074 China Hubei Key Laboratory of Engineering Modeling and Scientific Computing Huazhong University of Science and Technology Wuhan430074 China School of Mathematics and Statistics Yunnan University Kunming650500 China Academy of Mathematics and Systems Science Chinese Academy of Sciences Beijing100190 China School of Mathematical Sciences University of Chinese Academy of Sciences Beijing100049 China

We utilize the weak convergence method to establish the Freidlin-Wentzell large deviations principle (LDP) for stochastic delay differential equations (SDDEs) with super-linearly growing coefficients, which covers a large class of cases with non-globally Lipschitz coefficients. The key ingredient in our proof is the uniform moment estimate of the controlled equation, where we handle the super-linear growth of the coefficients by an iterative argument. Our results allow both the drift and diffusion coefficients of the considered equations to super-linearly grow not only with respect to the delay variable but also to the state variable. This work extends the existing results which develop the LDPs for SDDEs with super-linearly growing coefficients only with respect to the delay *** Codes 60H10, 60F10, 60H30 Copyright © 2022, The Authors. All rights reserved.

关键词： Stochastic systems

来源：评论

学校读者我要写书评

暂无评论

An Empirical Study of Data Sampling Techniques for Just-in-Time Software Defect Prediction

SSRN

引用

SSRN 2023年

作者： Li, Zhiqiang Du, Qiannan Niu, Jingwen Zhang, Hongyu Jing, Xiao-Yuan School of Computer Science Shaanxi Normal University Shaanxi Xi’an710119 China School of Computer and Information Engineering Xinxiang University Henan Xinxiang453003 China School of Electrical Engineering and Computing The University of Newcastle CallaghanNSW2308 Australia School of Computer Science Wuhan University Hubei Wuhan430072 China School of Computer Science Guangdong Provincial Key Laboratory on Petrochemical Equipment Fault Diagnosis Guangdong University of Petrochemical Technology Guangdong Maoming525000 China

Predicting whether there are defects in code changes submitted by developers is called just-in-time software defect prediction (JIT-SDP). Unfortunately, JIT-SDP usually suffers from the class imbalance problem, which affects its prediction performance. To overcome this problem, data sampling is one of the commonly-used class imbalance techniques. However, there is a lack of comprehensive empirical study to compare different data sampling techniques on the performance of JIT-SDP. In this paper, we consider both defect classification and defect ranking, two typical application scenarios. We perform an empirical comparison of 10 data sampling algorithms on the performance of JIT-SDP with three widely-used classifiers. Extensive experiments on 10 projects show that: (1) In the scenario of defect classification, sampling techniques can significantly improve the performance of Recall while are unfavorable for the performance of Precision and Pf. (2) In the scenario of defect ranking, sampling techniques can significantly improve the performance of Recall@20% while are unfavorable for the performance of Precision@20% and PCI@20%. Generally, ROM and SMOTE are more stable and effective sampling algorithms across several performance measures for JIT-SDP in both defect classification and defect ranking scenarios. Hence, our study suggests that future JIT-SDP research can use ROM and SMOTE rebalancing defect data. © 2023, The Authors. All rights reserved.

关键词： Defects

来源：评论

学校读者我要写书评

暂无评论

PointCA: Evaluating the Robustness of 3D Point Cloud Completion Models Against Adversarial Examples

arXiv

引用

arXiv 2022年

作者： Hu, Shengshan Zhang, Junwei Liu, Wei Hou, Junhui Li, Minghui Zhang, Leo Yu Jin, Hai Sun, Lichao School of Cyber Science and Engineering Huazhong University of Science and Technology China School of Computer Science and Technology Huazhong University of Science and Technology China School of Software Engineering Huazhong University of Science and Technology China Cluster and Grid Computing Lab China National Engineering Research Center for Big Data Technology and System Hubei Engineering Research Center on Big Data Security China Hubei Key Laboratory of Distributed System Security China Services Computing Technology and System Lab China City University of Hong Kong Hong Kong Deakin University Australia Lehigh University United States

Point cloud completion, as the upstream procedure of 3D recognition and segmentation, has become an essential part of many tasks such as navigation and scene understanding. While various point cloud completion models have demonstrated their powerful capabilities, their robustness against adversarial attacks, which have been proven to be fatally malicious towards deep neural networks, remains unknown. In addition, existing attack approaches towards point cloud classifiers cannot be applied to the completion models due to different output forms and attack purposes. In order to evaluate the robustness of the completion models, we propose PointCA, the first adversarial attack against 3D point cloud completion models. PointCA can generate adversarial point clouds that maintain high similarity with the original ones, while being completed as another object with totally different semantic information. Specifically, we minimize the representation discrepancy between the adversarial example and the target point set to jointly explore the adversarial point clouds in the geometry space and the feature space. Furthermore, to launch a stealthier attack, we innovatively employ the neighbourhood density information to tailor the perturbation constraint, leading to geometry-aware and distribution-adaptive modification for each point. Extensive experiments against different premier point cloud completion networks show that PointCA can cause a performance degradation from 77.9% to 16.7%, with the structure chamfer distance kept below 0.01. We conclude that existing completion models are severely vulnerable to adversarial examples, and state-of-the-art defenses for point cloud classification will be partially invalid when applied to incomplete and uneven point cloud data. Copyright © 2022, The Authors. All rights reserved.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Revisiting Token Dropping Strategy in Efficient BERT Pretraining

arXiv

引用

arXiv 2023年

作者： Zhong, Qihuang Ding, Liang Liu, Juhua Liu, Xuebo Zhang, Min Du, Bo Tao, Dacheng National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence School of Computer Science Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University China JD Explore Academy China Research Center for Graphic Communication Printing and Packaging Institute of Artificial Intelligence Wuhan University China Institute of Computing and Intelligence Harbin Institute of Technology China University of Sydney Australia

Token dropping is a recently-proposed strategy to speed up the pretraining of masked language models, such as BERT, by skipping the computation of a subset of the input tokens at several middle layers. It can effectively reduce the training time without degrading much performance on downstream tasks. However, we empirically find that token dropping is prone to a semantic loss problem and falls short in handling semantic-intense tasks (§2). Motivated by this, we propose a simple yet effective semantic-consistent learning method (SCTD) to improve the token dropping. SCTD aims to encourage the model to learn how to preserve the semantic information in the representation space. Extensive experiments on 12 tasks show that, with the help of our SCTD, token dropping can achieve consistent and significant performance gains across all task types and model sizes. More encouragingly, SCTD saves up to 57% of pretraining time and brings up to +1.56% average improvement over the vanilla token dropping. Copyright © 2023, The Authors. All rights reserved.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

ON COMPUTABILITY OF EQUILIBRIUM STATES

arXiv

引用

arXiv 2023年

作者： Binder, Ilia He, Qiandu Li, Zhiqiang Zhang, Yiwei Department of Mathematics University of Toronto Bahen Centre 40 St. George St. TorontoONM5S 2E4 Canada School of Mathematical Sciences Peking University Beijing100871 China School of Mathematical Sciences Beijing International Center for Mathematical Research Peking University Beijing100871 China School of Mathematics and Statistics Center for Mathematical Sciences Hubei Key Laboratory of Engineering Modeling and Scientific Computing Huazhong University of Science and Technology Wuhan430074 China

Equilibrium states are natural dynamical analogs of Gibbs states in thermodynamic formalism. This paper is devoted to the study of their computability in the sense of Computable Analysis. We show that the unique equilibrium state associated to a pair of a computable, topologically exact, distance-expanding, open map T: X → X and a computable Hölder continuous potential : X → R is always computable. Furthermore, the Hausdorff dimension of the Julia set and the equilibrium state for the geometric potential of a computable hyperbolic rational map are computable. On the other hand, we introduce a mechanism to establish non-uniqueness of equilibrium states. We also present some computable dynamical systems whose equilibrium states are all *** Codes Primary: 03D78, Secondary: 37D35, 37D20, 37F15 Copyright © 2023, The Authors. All rights reserved.

关键词： Dynamical systems

来源：评论

学校读者我要写书评

暂无评论

A new method for vascular age estimation based on relative risk difference in vascular aging

引用

Computers in Biology and Medicine 2024年 171卷 108155-108155页

作者： Tang, Qingfeng Liu, Shiping Tao, Chao Wang, Jue Zhao, Huanhuan Wang, Guangjun Zhao, Xu Ren, Qun Zhang, Liangliang Su, Benyue Xu, Jiatuo An, Hui School of Traditional Chinese Medicine Shanghai University of Traditional Chinese Medicine Shanghai201203 China The University Key Laboratory of Intelligent Perception and Computing of Anhui Province Anqing Normal University Anqing246133 China Anhui Engineering Research Center of Intelligent Perception and Elderly Care Chuzhou University Chuzhou239000 China School of Computer and Information Engineering Chuzhou University Chuzhou239000 China Health Management & Physical Examination Center Xiangyang Central Hospital Affiliated Hospital of Hubei University of Arts and Science Xiangyang441021 China School of Mathematics and Computer Science Tongling University Tongling244061 China

Objective: The current models of estimating vascular age (VA) primarily rely on the regression label expressed with chronological age (CA), which does not account individual differences in vascular aging (IDVA) that are difficult to describe by CA. This may lead to inaccuracies in assessing the risk of cardiovascular disease based on VA. To address this limitation, this work aims to develop a new method for estimating VA by considering IDVA. This method will provide a more accurate assessment of cardiovascular disease risk. Methods: Relative risk difference in vascular aging (RRDVA) is proposed to replace IDVA, which is represented as the numerical difference between individual predicted age (PA) and the corresponding mean PA of healthy population. RRDVA and CA are regard as the influence factors to acquire VA. In order to acquire PA of all samples, this work takes CA as the dependent variable, and mines the two most representative indicators from arteriosclerosis data as the independent variables, to establish a regression model for obtaining PA. Results: The proposed VA based on RRDVA is significantly correlated with 27 indirect indicators for vascular aging evaluation. Moreover, VA is better than CA by comparing the correlation coefficients between VA, CA and 27 indirect indicators, and RRDVA greater than zero presents a higher risk of disease. Conclusion: The proposed VA overcomes the limitation of CA in characterizing IDVA, which may help young groups with high disease risk to promote healthy behaviors. © 2024 Elsevier Ltd

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

SCARE of Secret Ciphers under Rough Leakage Model

引用

Security and Communication Networks 2023年第1期2023卷

作者： Wang, Mengxing Tang, Ming Zhang, Xige Chen, Qi Ding, Haili Key Laboratory of Aerospace Information Security and Trusted Computing Ministry of Education School of Cyber Science and Engineering Wuhan University Hubei Wuhan430000 China Beijing Smartchip Microelectronics Technology Co. Ltd. Beijing100192 China Marketing Service Center State Grid Electric Power Co. Ltd. Ningxia750002 China

Most previousside-channel analysis-based reverse engineering (SCARE) methods were based on collision attack. However, the collision-based methods generally need noise-free traces and can hardly be verified by realistic implementations. This paper proposes a novel SCARE, which is named as OSHDG-SCARE, targeting on secret S-box. We define a graph as OSHDG (one-step HD graph) to recover the parameters of any secret S-box. Due to the redundancy of OSHDG, the multiple paths can filter the correct results and improve the success rate of the reverse method even when the samples deviate from leakage model. We classify the deviation of side channel signal as two types, which are the gap between real leakage and leakage model, and measurement noise. Experiments are performed on real power traces satisfying the inaccurate HW (Hamming Weight) leakage model from a software implementation of AES-like cipher. OSHDG-SCARE recovers the secret S-box when the deviation rate of traces is 1.45%. © 2023 Mengxing Wang et al.

关键词： Side channel attack

来源：评论

学校读者我要写书评

暂无评论

An Auto-Parallel Method for Deep Learning Models Based on Genetic Algorithm

An Auto-Parallel Method for Deep Learning Models Based on Ge...

引用

International Conference on Parallel and Distributed Systems (ICPADS)

作者： Yan Zeng ChengChuang Huang YiJie Ni ChunBao Zhou JiLin Zhang Jue Wang MingYao Zhou MeiTing Xue YunQuan Zhang School of Computer Science and Technology Hangzhou Dianzi University Hangzhou China Key Laboratory for Modeling and Simulation of Complex Systems Ministry of Education Hangzhou China Data Security Governance Zhejiang Engineering Research Center Hangzhou China School of ITMO Joint Institute Hangzhou Dianzi University Hangzhou China Institute of Computer Network Information Center of the Chinese Academy of Sciences Beijing China HuaWei State Key Laboratory of Computer Architecture Institute of Computing Technology of the Chinese Academy of Sciences Beijing China

As the size of datasets and neural network models increases, automatic parallelization methods for models have become a research hotspot in recent years. The existing auto-parallel methods based on machine learning or graph algorithms still have issues with search efficiency and applicability. This paper proposes an automatic parallel method based on a dual-population genetic algorithm, TGA, which transforms model partitioning and placement into an integer linear programming problem and constructs a cost model to evaluate the solution. The solution space is built using the neural network’s dataflow graph and device cluster’s topology, and the dual-population genetic algorithm is used to search for the optimal model parallel strategy. Experiments with various models show that the proposed method can improve single-step execution time by up to 42% compared to the Baechi method and up to 37.7% compared to the Hierarchical method.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Three factors make bulk high-entropy alloys as effective electrocatalysts for oxygen evolution

材料展望（英文）

引用

材料展望（英文） 2023年第4期 110-120页

作者： Tao Zhang Hui-Feng Zhao Ke-Yan Wang Zhen-Jie Chen Li Li Jing Peng Xu Peng Yong-Jiang Huang Hai-Bin Yu Wuhan National High Magnetic Field Center & School of Physics Huazhong University of Science and TechnologyWuhan 430074People's Republic of China School of Materials Science and Engineering Center of Analysis Measurement and ComputingHarbin Institute of TechnologyHarbin 150001People's Republic of China Faculty of Materials Science and Energy Engineering/Institute of Technology for Carbon Neutrality Shenzhen 518055People's Republic of China Shenzhen Key Laboratory of Energy Materials for Carbon Neutrality Shenzhen Institutes of Advanced TechnologyChinese Academy of SciencesShenzhen 518055People's Republic of China College of Chemistry and Chemical Engineering Hubei UniversityWuhan 430062People's Republic of China

Even in their bulk forms,complex alloys like high-entropy alloys(HEAs)exhibit favorable activity and stability as electrocatalysts for the oxygen evolution reaction(OER).However,the underlying reasons are not yet fully *** a family of Mo-doped CrFeCoNi-based HEAs,we have identified three crucial factors that govern their performance:(ⅰ)homogeneous solid solution phase of HEAs helps to maintain high-valence states of metals;(ⅱ)surface reconstruction results in a hybrid material comprising amorphous domains and percolated crystalline structures;(ⅲ)diversity of active intermediate species(M-O,M-OOH,and,notably,the abundance of superoxide μ-OO),which display stronger adsorption capacity on the reconstructed *** results are revealing due to their resemblance to findings in other families of electrocatalysts for OER,as well as their unique features specific to *** line with these factors,a CrFeCoNiMo0.2 bulk integrated electrode displays a low overpotential of 215 mV,rapid kinetics,and long-term stability of over 90 *** HEAs hold great potential for industrial applications.

关键词： high-entropy alloy solid solution electrocatalysis DFT calculation oxygen evolution reaction

来源：评论

学校读者我要写书评

暂无评论

Differentiating Malicious and Benign Android App Operations Using Second-Step Behavior Features

引用

Chinese Journal of Electronics 2019年第5期28卷 944-952页

作者： LI Pengwei FU Jianming XU Chao CHENG Binlin ZHANG Huanguo School of Information Engineering Nanjing Audit University School of Cyber Science and Engineering Wuhan University Key Laboratory of Aerospace Information Security and Trusted Computing Ministry of Education Wuhan University College of Computer Science and Technology Hubei Normal University

Security-sensitive operations in Android applications(apps for short) can either be benign or malicious. In this work, we introduce an approach of static program analysis that extracts "second-step behavior features", i.e., what was triggered by the security-sensitive operation, to assist app analysis in differentiating between malicious and benign operations. Firstly, we summarized the characteristics of malicious operations, such as spontaneity, independence, stealthiness and continuity,which can be used to classify the malicious operations and benign ones. Secondly, according to these characteristics,Second step behavior features(SSBFs for short) have been presented, including structural features and semantic features. Thirdly, an analysis prototype named SSdroid has been implemented to automatically extract SSBFs of security-sensitive operations. Finally, experiments on 9285 operations from both benign and malicious apps show that SSBFs are effective and usefulness. Our evaluation results suggest that the second-step behavior can greatly assist in Android malware detection.

关键词： Android malware Security-sensitive operations Static program analysis Classification

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：