An evaluation of quality and robustness of smoothed explanations

no code implementations • 29 Sep 2021 • Ahmad Ajalloeian, Seyed-Mohsen Moosavi-Dezfooli, Michalis Vlachos, Pascal Frossard

However, a combination of additive and non-additive attacks can still manipulate these explanations, which reveals shortcomings in their robustness properties.

On the Convergence of SGD with Biased Gradients

no code implementations • 31 Jul 2020 • Ahmad Ajalloeian, Sebastian U. Stich

We analyze the complexity of biased stochastic gradient methods (SGD), where individual updates are corrupted by deterministic, i. e. biased error terms.