Brandon Tran

TL;DR: Spectral signatures are identified as a new property of all known backdoor attacks, which allows tools from robust statistics to thwart the attacks and is demonstrated the efficacy of these signatures in detecting and removing poisoned examples on real image sets and state of the art neural network architectures.

TL;DR: In this article, the authors investigate the vulnerability of neural network-based classifiers to rotations and translations, and find that first-order methods cannot reliably find worst-case perturbations.

TL;DR: The authors demonstrate that adversarial examples can be directly attributed to the presence of non-robust features: features derived from patterns in the data distribution that are highly predictive, yet brittle and incomprehensible to humans.

TL;DR: In this paper, the authors identify a new property of all known backdoor attacks, which they call ''spectral signatures'' and utilize tools from robust statistics to thwart the attacks and demonstrate the efficacy of these signatures in detecting and removing poisoned examples on real image sets and state of the art neural network architectures.