检索结果-内蒙古大学图书馆

arXiv 2022年

作者： Fan, Mingyuan Guo, Wenzhong Yu, Shengxing Ying, Zuobin Liu, Ximeng College of Computer and Data Science Fuzhou University China Faculty of Data Science City University of Macau China School of Electronics Engineering and Computer Science Peking University China

Transferability of adversarial examples is of critical importance to launch black-box adversarial attacks, where attackers are only allowed to access the output of the target model. However, under such a challenging but practical setting, the crafted adversarial examples are always prone to overfitting to the proxy model employed, presenting poor transferability. In this paper, we suggest alleviating the overfitting issue from a novel perspective, i.e., designing a fitted model architecture. Specifically, delving the bottom of the cause of poor transferability, we arguably decompose and reconstruct the existing model architecture into an effective model architecture, namely multi-track model architecture (MMA). The adversarial examples crafted on the MMA can maximumly relieve the effect of model-specified features to it and toward the vulnerable directions adopted by diverse architectures. Extensive experimental evaluation demonstrates that the transferability of adversarial examples based on the MMA significantly surpass other state-of-the-art model architectures by up to 40% with comparable overhead. Copyright © 2022, The Authors. All rights reserved.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Low-rank matrix recovery with total generalized variation for defending adversarial examples

引用

Frontiers of Information Technology & Electronic Engineering 2024年第3期25卷 432-445页

作者： Wen LI Hengyou WANG Lianzhi HUO Qiang HE Linlin CHEN Zhiquan HE Wing W.Y.Ng School of Science Beijing University of Civil Engineering and ArchitectureBeijing 100044China School of Computer Science and Engineering South China University of TechnologyGuangzhou 510006China Aerospace Information Research Institute Chinese Academy of SciencesBeijing 100094China Guangdong Key Laboratory of Intelligent Information Processing Shenzhen 518060China Institute of Big Data Modeling and Technology Beijing University of Civil Engineering and ArchitectureBeijing 100044China

Low-rank matrix decomposition with first-order total variation(TV)regularization exhibits excellent performance in exploration of image *** advantage of its excellent performance in image denoising,we apply it to improve the robustness of deep neural ***,although TV regularization can improve the robustness of the model,it reduces the accuracy of normal samples due to its *** our work,we develop a new low-rank matrix recovery model,called LRTGV,which incorporates total generalized variation(TGV)regularization into the reweighted low-rank matrix recovery *** the proposed model,TGV is used to better reconstruct texture information without *** reweighted nuclear norm and Li-norm can enhance the global structure ***,the proposed LRTGV can destroy the structure of adversarial noise while re-enhancing the global structure and local texture of the *** solve the challenging optimal model issue,we propose an algorithm based on the alternating direction method of *** results show that the proposed algorithm has a certain defense capability against black-box attacks,and outperforms state-of-the-art low-rank matrix recovery methods in image restoration.

关键词： Total generalized variation Low-rank matrix Alternating direction method of multipliers Adversarial example

来源：评论

学校读者我要写书评

暂无评论

Differentially Private Deep Learning with Iterative Gradient Descent Optimization

引用

ACM/IMS Transactions on data science 2021年第4期2卷 1–27页

作者： Ding, Xiaofeng Chen, Lin Zhou, Pan Jiang, Wenbin Jin, Hai National Engineering Research Center for Big Data Technology and System Lab Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Wuhan430074 China Hubei Engineering Research Center on Big Data Security School of Cyber Science and Engineering Huazhong University of Science and Technology Wuhan430074 China

Deep learning has achieved great success in various areas and its success is closely linked to the availability of massive data. But in general, a large dataset could include sensitive data and therefore the model should have the capability to avoid privacy leakage. To achieve this aim, many works apply the famous privacy framework named differential privacy into deep learning to preserve privacy. In this article, we propose a novel perturbed iterative gradient descent optimization (PIGDO) algorithm and prove that this algorithm satisfies the differential privacy. Besides, we propose a modified moments accountant (MMA) method to conduct the privacy analysis and obtain a tighter bound of privacy loss compared with the original moments accountant method. A number of experiments demonstrate that our optimization algorithm can not only improve the model accuracy and training speed, but also achieve better privacy guarantees over the state-of-the-art algorithm while reaching the equivalent accuracy. We provide codes for all of our experiments in https://***/CGCL-codes/***. © 2022 Association for Computing Machinery.

关键词： Large dataset

来源：评论

学校读者我要写书评

暂无评论

Analysis and Detection of Fake News Using Machine Learning 3

Analysis and Detection of Fake News Using Machine Learning

引用

3rd International Conference on Artificial Intelligence For Internet of Things, AIIoT 2024

作者： Sathish Kumar, P. Suthanthiradevi, P. Arul Stephen, C. Ebenezer Abishek, B. Sivakumar, S. Mathiyarasu, M. Computer and Communication Engineering Rajalakshmi institute of technology Chennai124 India Data Science and Business Systems School of Computing SRM institute of Science and Technology Kattankulathur Chennai603203 India Electronics and Communication Engineering Vels Institute of science Technology and Advanced studies Chennai117 India Electronics and Communication Engineering Vel Tech Multi Tech Dr. Rangarajan Dr. Sakunthala Engineering College Avadi Chennai600062 India Electronics and Communication Engineering BIHER Chennai73 India Computer and Communication Engineering Rajalakshmi institute of technology Chennai India

ISBN: (纸本)9798350372120

The Fake news becomes a more critical issue nowadays, it may reduce the trust of the public about the particular information. This project focuses on the advancement of fake news detection model using machine learning. The main goal of this project is to develop, analyze a model that can effectively classify the news articles based on their Subject or contents. Natural preprocessing Algorithm and features such as Stemming, Tokenization, Stop word and TF-IDF Vectorization are used to train the machine learning model. Then the Machine learning Algorithms likeLogistic Regression and Random Forest are used to classifying the dataset which gives 98% accuracy and 99% of accuracy score. For this Project we have taken a dataset of 22413 Real news and 23481 Fake News from the online resources. This model has a potential to reduce the impact of fake news by grant the promotion of accurate and reliable information in a society. © 2024 IEEE.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

A new method for a semi-positone Hadamard fractional boundary value problem

引用

Chaos, Solitons and Fractals: X 2024年 12卷

作者： Liu, Rui Zhai, Chengbo Ren, Jing School of Mathematical Sciences Shanxi University Shanxi Taiyuan030006 China Key Laboratory of Complex Systems and Data Science of Ministry of Education Shanxi University Shanxi Taiyuan030006 China School of Mathematics and Computer Science Shanxi Normal University Taiyuan030031 China

A Hadamard fractional boundary value problem with semi-positone nonlinearity is studied in this paper, and the local existence and uniqueness of solutions are derived by a recent fixed point theorem involving with increasing φ−(h,r)-concave operators defined on ordered spaces. This is an unprecedented approach to solve the semi-positone problems. In practice, this method has a wider range of applicability since it can obtain the local uniqueness of the solutions. Furthermore, we can approximate the unique solution by constructing convergent iterative sequences. In the end, a persuasive example is provided to illustrate that the theoretical results we obtained are applicable and valid. © 2023 The Author(s)

关键词： Boundary value problems

来源：评论

学校读者我要写书评

暂无评论

Semi-global Stereo Matching Algorithm Based on Census Transform 4

Semi-global Stereo Matching Algorithm Based on Census Transf...

引用

2022 4th International Conference on Electronic Engineering and Informatics, EEI 2022

作者： Hu, Jinjing Liang, Quan Yu, Wenze School of Computer Science and Mathematics Fujian University of Technology Fuzhou China Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of Technology Fuzhou China

ISBN: (纸本)9783800759231

Aiming at the problem that the center point of traditional Census transform window is vulnerable to the noise of outer boundary environment and the area of partial depth discontinuous area has low matching precision in stereo matching, a semi-global stereo matching algorithm based on anti-noise Census transform is proposed. Parallax graphs with a higher degree of match can be output. Firstly, by preprocessing the image, more information of the image can be retained and noise interference can be removed. Meanwhile, the Census algorithm is improved to overcome the dependence on the center pixel of the window and enhance the anti-interference of the initial generation value against noise. Secondly, in the cost aggregation stage, Sobel edge detection is used for cost aggregation. Then, the initial aberration is determined according to the rule that the winner is king. The last is the parallax optimization, through the left-right consistency detection strategy to reduce the false match points, improve the parallax accuracy of occlusion points, and finally get a fine parallax map. Compared with other stereo matching algorithms on Middlebury Vision website, the experimental results show that the proposed algorithm is superior to other stereo matching algorithms. © VDE VERLAG GMBH ∙ Berlin ∙ Offenbach.

关键词： Geometrical optics

来源：评论

学校读者我要写书评

暂无评论

Unsupervised Time Series Anomaly Detection under data Contamination 12

Unsupervised Time Series Anomaly Detection under Data Contam...

引用

12th International Conference on Information Technology in Medicine and Education, ITME 2022

作者： Lin, Xiaohui Li, Zuoyong Huang, Xunhua Chen, Xinwei Fan, Haoyi Fuzhou University College of Computer and Data Science Fuzhou350108 China Minjiang University Fujian Provincial Key Laboratory of Information Processing and Intelligent Control College of Computer and Control Engineering Fuzhou350121 China Harbin University of Science and Technology School of Computer Science and Technology Harbin150080 China Minjiang University Fujian Engineering and Research Center of New Chinese Lacquer Materials Fuzhou350121 China Zhengzhou University School of Computer and Artificial Intelligence Zhengzhou450001 China

ISBN: (纸本)9798350310153

Unsupervised learning utilizes unlabeled data to alleviate the reliance on large amounts of labeled data, and it has made great progress in time series anomaly detection. However, there are still some thorny issues unsolved, especially the incapability to maintain the robustness of the model when the time series dataset is contaminated. In this work, we propose an unsupervised time series anomaly detection method on data contamination, named distribution alignment autoencoder (DAAE), in which a temporal sliding reward function combines the adversarial distribution alignment to mine the contaminated data. Specifically, the dimensionality reduction and reconstruction of the time series are implemented by an autoencoder. Next, using two discriminators, the latent features obtained by dimensionality reduction are mapped to the Gaussian distribution, and the high-confidence normal time series are generated by sampling noise from the Gaussian distribution. Then, we compute the mean square error between high-confidence normal samples and the training data. Finally, we feed the error values to the temporal sliding reward function to mine the contaminated training samples. Experimental results compared with baselines on public datasets demonstrate the effectiveness of DAAE. © 2022 IEEE.

关键词： Contamination

来源：评论

学校读者我要写书评

暂无评论

Change of Electrical and Transport Properties of Nickel Oxide by Carrier Concentration and Temperature through First‑Principle Calculations

引用

Nanomanufacturing and Metrology 2023年第4期6卷 109-118页

作者： Nayem Md.Reza Shah Chang‑Dong Yeo Minyeong Choi Yang‑Ki Hong Jeong H.You Department of Mechanical Engineering Texas Tech UniversityLubbockTX 79409USA Department of Electrical and Computer Engineering and Material Science Ph.D.Program The University of AlabamaTuscaloosaAL 35487USA Department of Mechanical Engineering School of EngineeringUniversity of St.ThomasSt.PaulMN 55105USA

Nickel is typically used as one of the main components in electrical contact devices or *** oxide(NiO)is usually formed on the surfaces of electrodes and can negatively impact system performance by introducing electrical contact *** thermal,electrical,and transport properties of NiO,as a Mott insulator or a p-type semiconductor,can be altered by operating and environmental conditions such as temperature and stress/strain by *** this study,we inves-tigate the fundamental material properties of NiO through the first-principle ***,we obtain and compare the lattice parameter,magnetic moment,and electronic structure for NiO via the WIEN2K simulations with four different poten-tials(i.e.,GGA,GGA+U,LSDA,and LSDA+U).Then,using the WIEN2K simulation results with LSDA+U potential that produces a highly accurate bandgap for NiO,we calculate the electrical conductivity and electrical part of the thermal conductivity of nickel and NiO as a function of temperature and carrier concentration through the BoltzTraP *** simulation results revealed that the electrical conductivity relative to the relaxation time for NiO increases with the carrier concentration,while it shows a slightly decreasing trend with temperature under a fixed carrier *** contrast,the electrical part of the thermal conductivity shows an increasing trend considering carrier concentration and temperature.

关键词： Nickel oxide Magnetic moments Density of states Seebeck coefcient Electrical conductivity

来源：评论

学校读者我要写书评

暂无评论

Detection of TB from Chest X-ray: A Study with EfficientNet

Detection of TB from Chest X-ray: A Study with EfficientNet

引用

2023 International Conference on System, Computation, Automation and Networking, ICSCAN 2023

作者： Rama, A. Rajakumar, M.P. Mythili, N. Arunmozhi, S. Mohammed, Mazin Abed Rajinikanth, V. Saveetha School of Engineering Simats Department of CSE TN Chennai602105 India St.Joseph's College of Engineering Department of Artificial Intelligence and Data Science TN Chennai600119 India St.Joseph's Institute of Technology Department of Computer Science and Engineering TN Chennai600119 India Manakula Vinayagar Institute of Technology Department of Electronics and Communication Engineering TN Puducherry605 107 India College of Computer Science and Information Technology University of Anbar Department of Artificial Intelligence Anbar31001 Iraq Saveetha School of Engineering Simats Division of Research and Innovation Department of Computer Science and Engineering TN Chennai602105 India

ISBN: (纸本)9798350315127

The lung is one of the prime organs, and any disease in the lung causes mild to severe breathing problems;untreated lung disease will lead to several complications. Tuberculosis (TB) is a lung ailment that needs premature recognition and handling. The primary objective is to employ the deep-learning (DL) based TB detection using chest $X$-rays. Various stages of the proposed scheme consist of (i) data collection and resizing, (ii) DL-supported feature extraction, (iii) binary classification and five-fold cross-validation, and (iv) comparison with earlier results and confirming the merit of the scheme. This research implements EfficientNet (EN) variants to classify the chosen $\mathrm{X}$-rays into healthy/TB classes using the SoftMax classifier. The proposed scheme with EN_B2 (ENB2) has been successful in providing an accuracy of $96{\% }$ as far as detection accuracy is considered when compared to other methods. The superiority of the suggested strategy is also confirmed by an analysis using the most recent technology, which confirms the worth of the proposed system on the chosen $\mathrm{X}$-ray imagery. © 2023 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Radio Map Estimation via Latent-Domain Plug-and-Play Denoisers

Radio Map Estimation via Latent-Domain Plug-and-Play Denoise...

引用

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

作者： Le Xu Lei Cheng Junting Chen Wenqiang Pu Xiao Fu School of Electrical Engineering and Computer Science Oregon State University Corvallis USA College of Information Science and Electronic Engineering Zhejiang University Hangzhou China School of Science and Engineering The Chinese University of Hong Kong Shenzhen China Shenzhen Research Institute of Big Data Shenzhen China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Radio map estimation (RME) aims to construct a map of radio strength across multiple domains (e.g., space and frequency) from limited measurements. data-driven deep neural model-based RME showed promising performance, yet requiring excessive training resources. This work puts forth an RME approach that can effectively incorporate learned information without training over radio map data. Our idea is to employ the plug-and-play (PnP) denoising scheme from computational imaging. The PnP framework allows incorporating denoisers trained over natural images to handle other types of data, e.g., ocean sound fields and medical images, due to the similarity of their denoising processes. Conventional PnP methods mostly use the learned denoisers in the data domain. Instead, the proposed approach applies PnP in the latent domain through a tailored algorithm design and spatial-spectral factorization of radio maps. This way, the proposed RME method exhibits enhanced scalability and noise robustness. Simulations are used to illustrate the effectiveness of the proposed approach.

关键词： Training Computational modeling Scalability Oceans Signal processing algorithms Estimation Sea measurements Numerical simulation Robustness Speech processing

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：