Parsimonious Black-Box Adversarial Attacks via Efficient Combinatorial OptimizationDownload PDFOpen Website

2019 (modified: 11 Nov 2022)ICML 2019Readers: Everyone
Abstract: Solving for adversarial examples with projected gradient descent has been demonstrated to be highly effective in fooling the neural network based classifiers. However, in the black-box setting, the...
0 Replies

Loading