检索结果-内蒙古大学图书馆

RobustART: Benchmarking Robustness on Architecture Design and Training Techniques

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Tang, Shiyu Gong, Ruihao Wang, Yan Liu, Aishan Wang, Jiakai Chen, Xinyun Yu, Fengwei Liu, Xianglong Song, Dawn Yuille, Alan Torr, Philip H.S. Tao, Dacheng The State Key Lab of Software Development Environment Beihang University Beijing100191 China The SenseTime Beijing100191 China The Berkeley Artificial Intelligence Research Lab University of California Berkeley BerkeleyCA94701 United States The Department of Cognitive Science and Computer Science Johns Hopkins University BaltimoreMD21218 United States The Department of Engineering Science University of Oxford OxfordOX1 3PJ United Kingdom The JD Explore Academy Beijing101111 China

Deep neural networks (DNNs) are vulnerable to adversarial noises, which motivates the benchmark of model robustness. Existing benchmarks mainly focus on evaluating defenses, but there are no comprehensive studies of how architecture design and training techniques affect robustness. Comprehensively benchmarking their relationships is beneficial for better understanding and developing robust DNNs. Thus, we propose RobustART, the first comprehensive Robustness investigation benchmark on ImageNet regarding ARchitecture design (49 human-designed off-the-shelf architectures and 1200+ networks from neural architecture search) and Training techniques (10+ techniques, e.g., data augmentation) towards diverse noises (adversarial, natural, and system noises). Extensive experiments substantiated several insights for the first time, e.g., (1) adversarial training is effective for the robustness against all noises types for Transformers and MLP-Mixers;(2) given comparable model sizes and aligned training settings, CNNs > Transformers > MLP-Mixers on robustness against natural and system noises;Transformers > MLP-Mixers > CNNs on adversarial robustness;(3) for some light-weight architectures, increasing model sizes or using extra data cannot improve robustness. Our benchmark http://***/presents: (1) an open-source platform for comprehensive robustness evaluation;(2) a variety of pre-trained models to facilitate robustness evaluation;and (3) a new view to better understand the mechanism towards designing robust DNNs. We will continuously develop to this ecosystem for the community. Copyright © 2021, The Authors. All rights reserved.

关键词： computer vision

Verifiable Privacy Preservation Scheme for Outsourcing Medi-cal Image to Cloud Through ROI Based Crypto-Watermarking

学校读者我要写书评

暂无评论

Verifiable Privacy Preservation Scheme for Outsourcing Medi-...

IEEE/ACM Int'l Conference on & Int'l Conference on Cyber, Physical and Social Computing (CPSCom) Green Computing and Communications (GreenCom)

作者： Chuan Zhou Yi Zhou XingHan An Yan Liu Min Wang XiangZhi Liu Tianjin University Tianjin China Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China College of Environmental Science and Engineering Beijing Forestry University Beijing China Inspur Software Co. Ltd. Jinan China The Party School of cpc Jinan Municipal Committee Jinan China

A novel verifiable privacy preservation scheme for outsourcing medical image to cloud through ROI based crypto-watermarking is proposed in the paper. In the proposed scheme, data owner firstly carries out substitution of S-box for the region of interest (ROI) of medical image, and then separates the image into 4 most significant bits (MSBs) plane and 4 least significant bits (LSBs) plane images. Secondly, the hash value of ROI is embedded into the two separated bit plane images using reversible watermarking algorithm. Lastly, some selected hash values are transformed into the initial parameters of chaotic maps, and the two sharing secrets, which are produced through chaos based encryption algorithm, are finally outsourced to two different cloud servers. For authorized users, they can get shares from different cloud servers, and then can losslessly recover the original medical image through a series of decryption operations and extraction of watermarking. Furthermore, the users can verify whether the original image is completely reconstructed or not, they even can locate the tampered parts inside ROI if anyone of the sharing secrets is damaged. Some experiments analyses and comparisons are given to show the security and effectiveness of proposed scheme.

关键词：

Ai outperformed every dermatologist: Improved dermoscopic melanoma diagnosis through customizing batch logic and loss function in an optimized deep cnn architecture

学校读者我要写书评

暂无评论

arXiv 2020年

作者： PHAM, CONG TRI LUONG, MAI CHI VAN HOANG, DUNG DOUCET, ANTOINE School of Computer Science and Engineering Thuyloi University 175 Tay Son Dong Da Hanoi Viet Nam ICTLab University of Science and Technology of Hanoi Vietnam Academy of Science and Technology 18 Hoang Quoc Viet Cau Giay Hanoi Viet Nam FPT Software 17 Duy Tan Street Cau Giay district Ha Noi Viet Nam Institute of Information Technology Vietnam Academy of Science and Technology 18 Hoang Quoc Viet Cau Giay Hanoi Viet Nam QuangBinh University Dong Hoi Quang Binh Viet Nam Laboratory L3i University La Rochelle France

Melanoma, one of most dangerous types of skin cancer, re-sults in a very high mortality rate. Early detection and resection are two key points for a successful cure. Recent research has used artificial in-telligence to classify melanoma and nevus and to compare the assess-ment of these algorithms to that of dermatologists. However, an imbal-ance of sensitivity and specificity measures affected the performance of existing models. This study proposes a method using deep convolution-al neural networks aiming to detect melanoma as a binary classification problem. It involves 3 key features, namely customized batch logic, customized loss function and reformed fully connected layers. The training dataset is kept up to date including 17,302 images of melanoma and nevus;this is the largest dataset by far. The model performance is compared to that of 157 dermatologists from 12 university hospitals in Germany based on MClass-D dataset. The model outperformed all 157 dermatologists and achieved state-of-the-art performance with AUC at 94.4% with sensitivity of 85.0% and specificity of 95.0% using a pre-diction threshold of 0.5 on the MClass-D dataset of 100 dermoscopic images. Moreover, a threshold of 0.40858 showed the most balanced measure compared to other researches, and is promisingly application to medical diagnosis, with sensitivity of 90.0% and specificity of 93.8%. Copyright © 2020, The Authors. All rights reserved.

关键词： Dermatology

Fair and autonomous sharing of federate learning models in mobile Internet of Things

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Hao, Xiaohan Ren, Wei Xiong, Ruoting Zheng, Xianghan Zhu, Tianqing Xiong, Neal N. School of Computer Science China University of Geoscience Wuhan China Guangxi Key Laboratory of Cryptography and Information Security Guilin541004 China Key Laboratory of Network Assessment Technology CAS Institute of Information Engineering Chinese Academy of Sciences Beijing100093 China Fujian Key Laboratory of Network Computing and Intelligent Information Processing Fuzhou University Fuzhou China Mingbyte Technology QingDao China School of Software University of Technology Sydney UltimoNSW2007 Australia Dept. of Mathematics and Computer Science Northeastern State University United States

Federate learning can conduct machine learning as well as protect the privacy of self-owned training data on corresponding ends, instead of having to upload to a central trusted data aggregation server. In mobile scenarios, a centralized trusted server may not be existing, and even though it exists, the delay will not be manageable, e.g., smart driving cars. Thus, mobile federate learning at the edge with privacy-awareness is attracted more and more attentions. It then imposes a problem - after data are trained on a mobile terminal to obtain a learned model, how to share the model parameters among others to create more accurate and robust accumulative final model. This kind of model sharing confronts several challenges, e.g., the sharing must be conducted without a third trusted party (autonomously), and the sharing must be fair as model training (by training data)is valuable. To tackle the above challenges, we propose a smart contract and IPFS (Inter-Planetary File System) based model sharing protocol and algorithms to address the challenges. The proposed protocol does not rely on a trusted third party, where individual-learned models are shared/stored in corresponding ends. Conducted through extensive experiments, three main steps of the proposed protocol are evaluated. The average executive time of the three steps are 0.059s, 0.060s and 0.032s, demonstrating its efficiency. Copyright © 2020, The Authors. All rights reserved.

关键词： Smart contract

Study on mechanisms of secure data aggregation based on weighted fitting analysis

学校读者我要写书评

暂无评论

Ruan Jian Xue Bao/Journal of software 2013年第SUPPL.1期24卷 108-116页

作者： Li, Ping Yang, Wu Xie, Jin-Yang Zhu, Hong-Song Zhang, Yong-Guang Li, Xiao-Feng School of Computer and Telecommunications Changsha University of Science and Technology Changsha 410114 China State Key Laboratory of Information Security Institute of Information Engineering The Chinese Academy of Sciences Beijing 100093 China Software Development Center Beijing Institute of Control Engineering Beijing 100190 China

Cryptographic approaches are vulnerable to impersonation attacks when nodes become compromised. Meanwhile, trust-based judgment and confirmation of cluster heads in WSN imposes higher communication cost. Trust management and authentication on cluster heads in WSN are essential in research on the mechanisms of secure data aggregation at present. In this paper, an energy attenuation model is adopted to describe event sensing in a kind of WSNs. Approximate evaluation on distance between occurring event and nodes within a cluster is achieved in case the location of the event is not known. Distribution properties of data points with clustering characteristics based on fitting curve construction are theoretically analyzed. In addition, the mechanism on trust authentication on cluster heads is also refined. Experimental results show that the proposed mechanism has much better performance on fitting accuracy and malicious CH recognition than previous works. © 2013 ISCAS.

关键词： Authentication

Joint learning of discriminative low-dimensional image representations based on dictionary learning and two-layer orthogonal projections

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Wei, Xian Shen, Hao Li, Yuanxiang Tang, Xuan Jin, Bo Zhao, Lijun Murphey, Yi Lu Fujian Institute of Research on the Structure of Matter Chinese Academy of Sciences China Technical University of Munich Germany and fortiss GmbH Munich Germany Shanghai Key Lab for Trustworthy Computing School of Computer Science and Software Engineering East China Normal University China School of Aeronautics & Astronautics Shanghai Jiao Tong University Shanghai200240 China State Key Laboratory of Robotics and System Harbin Institute of Technology Harbin150006 China Department of Electrical and Computer Engineering University of Michigan-Dearborn DearbornMI48128 United States

This work investigates the problem of efficiently learning discriminative low-dimensional representations of multi-class large-scale image objects. We propose a generic deep learning approach by taking advantages of Convolutional Neural Networks (CNN), sparse dictionary learning, and orthogonal projections. CNN is not only powerful on feature extraction, but also robust to spatial variance and changes. Sparse dictionary learning is well known for disentangling nonlinear underlying discriminative factors in data. The orthogonal projection is a notable efficient tool to project multi-class data onto low dimensional discriminative subspace. The proposed procedure can be summarized as follows. At first, a CNN is employed to extract high-dimensional (HD) preliminary convolutional features. Secondly, to avoid the high computational cost of direct sparse coding on HD CNN features, we present to learn sparse representation (SR) in an orthogonal projected space over a task-driven sparsifying dictionary. We then exploit the discriminative projection on SR. The whole learning process is treated as a joint optimization problem of trace quotient maximization, which involves the CNN parameters, the orthogonal projection on CNN features, the dictionary and the discriminative projection on sparse codes. The related cost function is well defined on product manifold of the Stiefel manifold, the Oblique manifold, and the Grassmann manifold. It is optimized via a geometrical stochastic gradient descent algorithm. Finally, the quality of dictionary and projections learned by the proposed approach is further investigated in terms of image classification. The experimental results show that the approach can achieve a strongly competitive performance with state-of-the-art image classification methods. Copyright © 2019, The Authors. All rights reserved.

关键词： Convolution

BG-GAN: Generative AI Enable Representing Brain Structure-Function Connections for Alzheimer’s Disease

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zhou, Tong Ding, Chen Jing, Changhong Liu, Feng Hung, Kevin Pham, Hieu Mahmud, Mufti Lyu, Zhihan Qiao, Sibo Wang, Shuqiang Tsang, Kim-Fung The Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences Shenzhen518055 China University of Chinese Academy of Sciences Beijing100864 China The Department of Systems and Enterprises Stevens Institute of Technology United States School of Science and Technology Hong Kong Metropolitan University Hong Kong College of Engineering and Computer Science VinUni-Illinois Smart Health Center VinUniversity Viet Nam Department of Information and Computer Science SDAIA-KFUPM Joint Research Center for AI Interdisciplinary Research Center for Biosystems and Machines King Fahd University of Petroleum and Minerals Dhahran Saudi Arabia Department of Game Design Faculty of Arts Uppsala University Sweden The School of Software Tiangong University Tianjin300387 China

The relationship between brain structure and function is critical for revealing the pathogenesis of brain disorders, including Alzheimer’s disease (AD). However, mapping brain structure to function connections is a very challenging task. In this work, a bidirectional graph generative adversarial network (BG-GAN) is proposed to represent brain structure-function connections. Specifically, by designing a module incorporating inner graph convolution network (InnerGCN), the generators of BG-GAN can employ features of direct and indirect brain regions to learn the mapping function between the structural domain and the functional domain. Besides, a new module named Balancer is designed to counterpoise the optimization between generators and discriminators. By introducing the Balancer into BG-GAN, both the structural generator and functional generator can not only alleviate the issue of mode collapse but also learn complementarity of structural and functional features. Experimental results using Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset show that both generated structure and function connections can improve the identification accuracy of AD. The experimental findings suggest that the relationship between brain structure and function is not a complete one-to-one correspondence. They also suggest that brain structure is the basis of brain function, and the strong structural connections are majorly accompanied by strong functional connections. Copyright © 2023, The Authors. All rights reserved.

关键词： Neurodegenerative diseases

Multi-error-correcting amplitude damping codes

学校读者我要写书评

暂无评论

Multi-error-correcting amplitude damping codes

IEEE International Symposium on Information Theory

作者： Runyao Duan Markus Grassl Zhengfeng Ji Bei Zeng State Key Laboratory of Intelligent Technology and Systems Tsinghua National Laboratory for Information Science and Technology Department of Computer Science and Technology Tsinghua University Beijing China Centre for Quantum Computation and Intelligent Systems (QCIS) Faculty of Engineering and Information Technology University of Technology Sydney NSW Australia Centre for Quantum Technologies National University of Singapore Singapore Perimeter Institute of Theoretical Physics Waterloo ONT Canada State Key Laboratory of Computer Science Institute of Software Chinese Academy and Sciences Beijing China Institute for Quantum Computing and the Department of Combinatorics and Optimization University of Waterloo Waterloo ONT Canada

ISBN: (纸本)9781424478903

We construct new families of multi-error-correcting quantum codes for the amplitude damping channel. Our key observation is that, with proper encoding, two uses of the amplitude damping channel simulate a quantum erasure channel. This allows us to use concatenated codes with quantum erasure-correcting codes as outer codes for correcting multiple amplitude damping errors. Our new codes are degenerate stabilizer codes and have parameters which are better than the amplitude damping codes obtained by any previously known construction.

关键词： Damping Quantum mechanics Quantum computing Error correction codes Noise level Laboratories Phase noise Physics computing Intelligent systems computer science

RETRACTED ARTICLE: Physically-based modeling and animation of tree branch patterns

学校读者我要写书评

暂无评论

The Visual computer 2009年第4期26卷 293-293页

作者： Yang, Meng Sheng, Bin Wu, Enhua Sun, Hanqiu Inst. of Software Chinese Academy of Sciences Beijing China Department of Computer Science and Engineering The Chinese University of Hong Kong Hong Kong China University of Macau & Inst. of Software Chinese Academy of Sciences Macau China