1 code implementation • 26 Jul 2021 • Alfred Laugros, Alice Caplier, Matthieu Ospici
Using the overlapping criterion, we split synthetic corruptions into categories that help to better understand neural network robustness.
no code implementations • 26 May 2021 • Alfred Laugros, Alice Caplier, Matthieu Ospici
To estimate the robustness of neural networks to these common corruptions, we generally use a group of modeled corruptions gathered into a benchmark.
no code implementations • 1 Jan 2021 • Alfred Laugros, Alice Caplier, Matthieu Ospici
In this paper, we propose to build corruption benchmarks with only non-overlapping corruptions, to improve their coverage and their balance.
no code implementations • 19 Aug 2020 • Alfred Laugros, Alice Caplier, Matthieu Ospici
Despite their performance, Artificial Neural Networks are not reliable enough for most of industrial applications.
no code implementations • 4 Sep 2019 • Alfred Laugros, Alice Caplier, Matthieu Ospici
We intend to study the links between the robustnesses of neural networks to both perturbations.