|
|
|
|
LEADER |
01529 am a22002053u 4500 |
001 |
137450 |
042 |
|
|
|a dc
|
100 |
1 |
0 |
|a Weng, Tsui-Wei
|e author
|
700 |
1 |
0 |
|a Zhang, Huan
|e author
|
700 |
1 |
0 |
|a Chen, Pin-Yu
|e author
|
700 |
1 |
0 |
|a Lozano, Aurelie
|e author
|
700 |
1 |
0 |
|a Hsieh, Cho-Jui
|e author
|
700 |
1 |
0 |
|a Daniel, Luca
|e author
|
245 |
0 |
0 |
|a ON EXTENSIONS OF CLEVER: A NEURAL NETWORK ROBUSTNESS EVALUATION ALGORITHM
|
260 |
|
|
|b IEEE,
|c 2021-11-05T13:37:18Z.
|
856 |
|
|
|z Get fulltext
|u https://hdl.handle.net/1721.1/137450
|
520 |
|
|
|a © 2018 IEEE. CLEVER (Cross-Lipschitz Extreme Value for nEtwork Robustness) is an Extreme Value Theory (EVT) based robustness score for large-scale deep neural networks (DNNs). In this paper, we propose two extensions on this robustness score. First, we provide a new formal robustness guarantee for classifier functions that are twice differentiable. We apply extreme value theory on the new formal robustness guarantee and the estimated robustness is called second-order CLEVER score. Second, we discuss how to handle gradient masking, a common defensive technique, using CLEVER with Backward Pass Differentiable Approximation (BPDA). With BPDA applied, CLEVER can evaluate the intrinsic robustness of neural networks of a broader class - networks with non-differentiable input transformations. We demonstrate the effectiveness of CLEVER with BPDA in experiments on a 121-layer Densenet model trained on the ImageNet dataset.
|
546 |
|
|
|a en
|
655 |
7 |
|
|a Article
|
773 |
|
|
|t 10.1109/globalsip.2018.8646356
|