Code implementation of the paper "Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks", at IEEE Security and Privacy 2019.
Primary LanguagePythonMIT LicenseMIT
No one’s star this repository yet.