发布求助
文献互助 智能选刊 最新文献

Journal of Computational Vision and Imaging Systems最新文献

筛选
英文 中文
On Robustness of Deep Neural Networks: A Comprehensive Study on the Effect of Architecture and Weight Initialization to Susceptibility and Transferability of Adversarial Attacks 深度神经网络的鲁棒性:体系结构和权重初始化对对抗性攻击敏感性和可转移性影响的综合研究
Journal of Computational Vision and Imaging Systems Pub Date : 2018-12-24 DOI: 10.15353/JCVIS.V4I1.329
Ibrahim Ben Daya, Mohammad Javad Shaifee, M. Karg, Christian Scharfenderger, A. Wong
{"title":"On Robustness of Deep Neural Networks: A Comprehensive Study on the Effect of Architecture and Weight Initialization to Susceptibility and Transferability of Adversarial Attacks","authors":"Ibrahim Ben Daya, Mohammad Javad Shaifee, M. Karg, Christian Scharfenderger, A. Wong","doi":"10.15353/JCVIS.V4I1.329","DOIUrl":"https://doi.org/10.15353/JCVIS.V4I1.329","url":null,"abstract":"Neural network models have shown state of the art performance inseveral applications. However it has been observed that they aresusceptible to adversarial attacks: small perturbations to the inputthat fool a network model into mislabelling the input data. Theseattacks can also transfer from one network model to another, whichraises concerns over their applicability, particularly when there areprivacy and security risks involved. In this work, we conduct a studyto analyze the effect of network architectures and weight initial-ization on the robustness of individual network models as well astransferability of adversarial attacks. Experimental results demon-strate that while weight initialization has no affect on the robustnessof a network model, it does have an affect on attack transferabilityto a network model. Results also show that the complexity of anetwork model as indicated by the total number of parameters andMAC number is not indicative of a network’s robustness to attackor transferability, but accuracy can be; within the same architec-ture, higher accuracy usually indicates a more robust network, butacross architectures there is no strong link between accuracy androbustness.","PeriodicalId":111292,"journal":{"name":"Journal of Computational Vision and Imaging Systems","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116964450","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 2
ConvART: Improving Adaptive Resonance Theory for Unsupervised Image Clustering 改进的自适应共振理论用于无监督图像聚类
Journal of Computational Vision and Imaging Systems Pub Date : 2018-12-24 DOI: 10.15353/JCVIS.V4I1.326
Ilia Sucholutsky, Matthias Schonlau
{"title":"ConvART: Improving Adaptive Resonance Theory for Unsupervised Image Clustering","authors":"Ilia Sucholutsky, Matthias Schonlau","doi":"10.15353/JCVIS.V4I1.326","DOIUrl":"https://doi.org/10.15353/JCVIS.V4I1.326","url":null,"abstract":"While supervised learning techniques have become increasinglyadept at separating images into different classes, these techniquesrequire large amounts of labelled data which may not always beavailable. We propose a novel neuro-dynamic method for unsuper-vised image clustering by combining 2 biologically-motivated mod-els: Adaptive Resonance Theory (ART) and Convolutional Neu-ral Networks (CNN). ART networks are unsupervised clustering al-gorithms that have high stability in preserving learned informationwhile quickly learning new information. Meanwhile, a major prop-erty of CNNs is their translation and distortion invariance, whichhas led to their success in the domain of vision problems. Byembedding convolutional layers into an ART network, the usefulproperties of both networks can be leveraged to identify differentclusters within unlabelled image datasets and classify images intothese clusters. In exploratory experiments, we demonstrate thatthis method greatly increases the performance of unsupervisedART networks on a benchmark image dataset.","PeriodicalId":111292,"journal":{"name":"Journal of Computational Vision and Imaging Systems","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124752623","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 1
MonolithNet: Training monolithic deep neural networks via a partitioned training strategy MonolithNet:通过分区训练策略训练单片深度神经网络
Journal of Computational Vision and Imaging Systems Pub Date : 2018-12-24 DOI: 10.15353/JCVIS.V4I1.340
Rene Bidart, A. Wong
{"title":"MonolithNet: Training monolithic deep neural networks via a partitioned training strategy","authors":"Rene Bidart, A. Wong","doi":"10.15353/JCVIS.V4I1.340","DOIUrl":"https://doi.org/10.15353/JCVIS.V4I1.340","url":null,"abstract":"In this study, we explore the training of monolithic deep neural net-works in an effective manner. One of the biggest challenges withtraining such networks to the desired level of accuracy is the dif-ficulty in converging to a good solution using iterative optimizationmethods such as stochastic gradient descent due to the enormousnumber of parameters that need to be learned. To achieve this,we introduce a partitioned training strategy, where proxy layersare connected to different partitions of a deep neural network toenable isolated training of a much smaller number of parametersto convergence. To illustrate the efficacy of this training strategy,we introduce MonolithNet, a massive residual deep neural networkconsisting of 437 million parameters. The trained MonolithNet wasable to achieve a top-1 accuracy of 97% on the CIFAR10 imageclassification dataset, which demonstrates the feasibility of the pro-posed training strategy for training monolithic deep neural networksto high accuracies.","PeriodicalId":111292,"journal":{"name":"Journal of Computational Vision and Imaging Systems","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116020660","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 1
Near-field Sensors with Machine Learning for Breast Tumor Detection 基于机器学习的近场传感器用于乳腺肿瘤检测
Journal of Computational Vision and Imaging Systems Pub Date : 2018-12-24 DOI: 10.15353/JCVIS.V4I1.328
M. Aldhaeebi, T. Almoneef, O. Ramahi
{"title":"Near-field Sensors with Machine Learning for Breast Tumor Detection","authors":"M. Aldhaeebi, T. Almoneef, O. Ramahi","doi":"10.15353/JCVIS.V4I1.328","DOIUrl":"https://doi.org/10.15353/JCVIS.V4I1.328","url":null,"abstract":"In this work, we propose the use of an electrically small novel antenna as a probe combined with a classification algorithm for nearfield microwave breast tumor detection. The resonant probe ishighly sensitive to the changes in the electromagnetic properties of the breast tissues such that the presence of the tumor is estimatedby determining the changes in the magnitude and phase responseof the reflection coefficient of the sensor. The Principle Component placed at the middle of the probe as shown in Fig. 1. The mainAnalysis (PCA) feature extraction method is applied to emphasize the difference in the probe responses for both the healthy and thetumourous cases . We show that when a numerical realistic breast with and without tumor cells is placed in the near field of the probe, the probe is capable of distinguishing between healthy and tumorous tissue. In addition, the probe is able to identify tumors with various sizes placed in single locations.","PeriodicalId":111292,"journal":{"name":"Journal of Computational Vision and Imaging Systems","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131221646","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Guarding Against Adversarial Attacks using Biologically Inspired Contour Integration 利用生物启发的轮廓整合来防范对抗性攻击
Journal of Computational Vision and Imaging Systems Pub Date : 2018-12-23 DOI: 10.15353/JCVIS.V4I1.336
Salman Hameed Khan, Alexander Wong, B. Tripp
{"title":"Guarding Against Adversarial Attacks using Biologically Inspired Contour Integration","authors":"Salman Hameed Khan, Alexander Wong, B. Tripp","doi":"10.15353/JCVIS.V4I1.336","DOIUrl":"https://doi.org/10.15353/JCVIS.V4I1.336","url":null,"abstract":"Artificial vision systems are susceptible to adversarial attacks. Smallintentional changes to images can cause these systems to mis-classify with high confidence. The brain has many mechanisms forstrengthening weak or confusing inputs. One such technique, con-tour integration can separate objects from irrelevant background.We show that incorporating contour integration within artificial vi-sual systems can increase their robustness to adversarial attacks.","PeriodicalId":111292,"journal":{"name":"Journal of Computational Vision and Imaging Systems","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121992263","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 1
SAMSON: Spectral Absorption-fluorescence Microscopy System for ON-site-imaging of algae SAMSON:用于藻类现场成像的光谱吸收荧光显微镜系统
Journal of Computational Vision and Imaging Systems Pub Date : 2018-10-02 DOI: 10.15353/jcvis.v4i1.324
Jason L. Deglint, L. Tang, Yitian Wang, Chao Jin, A. Wong
{"title":"SAMSON: Spectral Absorption-fluorescence Microscopy System for ON-site-imaging of algae","authors":"Jason L. Deglint, L. Tang, Yitian Wang, Chao Jin, A. Wong","doi":"10.15353/jcvis.v4i1.324","DOIUrl":"https://doi.org/10.15353/jcvis.v4i1.324","url":null,"abstract":"This paper presents SAMSON, a Spectral Absorption-fluorescenceMicroscopy System for ON-site-imaging of algae within a watersample. Designed to be portable and low-cost for on-site use,the optical sub-system of SAMSON consists of a mixture of low-cost optics and electronics, designed specifically to capture bothfluorescent and absorption responses from a water sample. Thegraphical user interface (GUI) sub-system of SAMSON was de-signed to enable flexible visualisation of algae in the water samplein real-time, with the ability to perform fine-grained exposure con-trol and illumination wavelength selection. We demonstrate SAM-SON’s capabilities by equipping the system with two fluorescentillumination sources and seven absorption illumination sources toenable the capture of multispectral data from six different algaespecies (three from the Cyanophyta phylum (blue-green algae) andthree from the Chlorophyta phylum (green algae)). The key benefitof SAMSON is the ability to perform rapid acquisition of fluores-cence and absorption data at different wavelengths and magnifica-tion levels, thus opening the door for machine learning methods toautomatically identify and enumerate different algae in water sam-ples using this rich wealth of data.","PeriodicalId":111292,"journal":{"name":"Journal of Computational Vision and Imaging Systems","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129372886","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 3
Lane Discovery in Traffic Video 交通视频中的车道发现
Journal of Computational Vision and Imaging Systems Pub Date : 1969-12-31 DOI: 10.15353/vsnl.v2i1.125
Nicholas Miller, David M. Swart, Ajay K. Mishra, Andrew Achkar
{"title":"Lane Discovery in Traffic Video","authors":"Nicholas Miller, David M. Swart, Ajay K. Mishra, Andrew Achkar","doi":"10.15353/vsnl.v2i1.125","DOIUrl":"https://doi.org/10.15353/vsnl.v2i1.125","url":null,"abstract":"Video sensing has become very important in Intelligent Transportation Systems (ITS) due to its relative low cost and non-invasive deployment. An effective ITS requires detailed traffic information, including vehicle volume counts for each lane in surveillance video of a highway or an intersection. The multiple-target, vehicle-tracking and counting problem is most reliably solved in a reduced space defined by the constraints of the vehicles driving within lanes. This requires lanes to be pre-specified. An off-line pre-processing method is presented which automatically discovers traffic lanes from vehicle motion in uncalibrated video from a stationary camera. A moving vehicle density map is constructed, then multiple lane curves are fitted. Traffic lanes are found without relying on possibly noisy tracked vehicle trajectories.","PeriodicalId":111292,"journal":{"name":"Journal of Computational Vision and Imaging Systems","volume":"107 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1969-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127757451","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整，为了账户安全，请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
请完成安全验证×
相关产品
×
本文献相关产品
联系我们：info@booksci.cn Book学术提供免费学术资源搜索服务，方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号：604180095
Book学术官方微信
小红书