no code implementations • 15 Jun 2023 • Feisi Fu, Wenchao Li
Existing ownership verification methods either modify or introduce constraints to the neural network parameters, which are accessible to an attacker in a white-box attack and can be harmful to the network's normal operation, or train the network to respond to specific watermarks in the inputs similar to data poisoning-based backdoor attacks, which are susceptible to backdoor removal techniques.
no code implementations • 2 Nov 2022 • Feisi Fu, Panagiota Kiourti, Wenchao Li
We present a novel methodology for neural network backdoor attacks.
no code implementations • 15 Aug 2022 • Zhilu Wang, YiXuan Wang, Feisi Fu, Ruochen Jiao, Chao Huang, Wenchao Li, Qi Zhu
Moreover, GROCET provides differentiable global robustness, which is leveraged in the training of globally robust neural networks.
2 code implementations • ICLR 2022 • Feisi Fu, Wenchao Li
By leveraging the piecewise linear nature of ReLU networks, our approach can efficiently construct a patch network tailored to the linear region where the buggy input resides, which when combined with the original network, provably corrects the behavior on the buggy input.