Xu, Peipei 
ORCID: 0000-0001-5866-2814, Ruan, Wenjie and Huang, Xiaowei ORCID: 0000-0001-6267-0366
(2022)
Quantifying safety risks of deep neural networks.
Complex and Intelligent Systems, 9 (4).
pp. 3801-3818.
|
Text
CIS_Springer2021.pdf - Author Accepted Manuscript Download (2MB) | Preview |
Abstract
<jats:title>Abstract</jats:title><jats:p>Safety concerns on the deep neural networks (DNNs) have been raised when they are applied to critical sectors. In this paper, we define safety risks by requesting the alignment of network’s decision with human perception. To enable a general methodology for quantifying safety risks, we define a generic safety property and instantiate it to express various safety risks. For the quantification of risks, we take the maximum radius of safe norm balls, in which no safety risk exists. The computation of the maximum safe radius is reduced to the computation of their respective Lipschitz metrics—the quantities to be computed. In addition to the known adversarial example, reachability example, and invariant example, in this paper, we identify a new class of risk—uncertainty example—on which humans can tell easily, but the network is unsure. We develop an algorithm, inspired by derivative-free optimization techniques and accelerated by tensor-based parallelization on GPUs, to support an efficient computation of the metrics. We perform evaluations on several benchmark neural networks, including ACSC-Xu, MNIST, CIFAR-10, and ImageNet networks. The experiments show that our method can achieve competitive performance on safety quantification in terms of the tightness and the efficiency of computation. Importantly, as a generic approach, our method can work with a broad class of safety risks and without restrictions on the structure of neural networks.</jats:p>
| Item Type: | Article |
|---|---|
| Uncontrolled Keywords: | Adversarial examples, Lipschitz metrics, Neural networks, Robustness, Safety, Uncertainty |
| Divisions: | Faculty of Science and Engineering > School of Electrical Engineering, Electronics and Computer Science |
| Depositing User: | Symplectic Admin |
| Date Deposited: | 24 Aug 2022 09:47 |
| Last Modified: | 07 Aug 2023 02:42 |
| DOI: | 10.1007/s40747-022-00790-x |
| Related URLs: | |
| URI: | https://livrepository.liverpool.ac.uk/id/eprint/3162120 |
Dimensions
Dimensions
