2018/02/05 by Alexey Chaplygin, Chaplygin, Alexey, Joshua Chacksfield +1
Computer Science · Mathematics · #Adversarial Robustness in Machine Learning #Anomaly Detection Techniques and Applications #Artificial Intelligence (cs.AI) #Artificial intelligence #Artificial neural network #Classifier (UML) #Computer Vision and Pattern Recognition (cs.CV) #Computer science #Contextual image classification #Convolutional neural network #Data mining #Deep learning #Digital Imaging for Blood Diseases #FOS: Computer and information sciences #Image (mathematics) #Image Retrieval and Classification Techniques #Machine Learning (stat.ML) #Machine Learning and Data Classification #Machine learning #Neural Networks and Applications #Pattern recognition (psychology) #Robustness (evolution) #Training set #cs.AI #cs.CV #stat.ML
paper · pdf · doi:10.48550/arxiv.1802.01435
9 pages
arxiv created 2018/02/05 · openalex publication_date 2018/02/05 · arxiv updated 2018/02/06 · openalex created_date 2022/10/01 · openalex updated_date 2026/08/05
Convolutional Neural Networks are a well-known staple of modern image\nclassification. However, it can be difficult to assess the quality and\nrobustness of such models. Deep models are known to perform well on a given\ntraining and estimation set, but can easily be fooled by data that is\nspecifically generated for the purpose. It has been shown that one can produce\nan artificial example that does not represent the desired class, but activates\nthe network in the desired way. This paper describes a new way of\nreconstructing a sample from the training set distribution of an image\nclassifier without deep knowledge about the underlying distribution. This\nenables access to the elements of images that most influence the decision of a\nconvolutional network and to extract meaningful information about the training\ndistribution.\n