1 code implementation • 9 Apr 2024 • Theo Di Piazza, Enric Meinhardt-Llopis, Gabriele Facciolo, Benedicte Bascle, Corentin Abgrall, Jean-Clement Devaux
Here, we demonstrate that the performance of these methods can be significantly enhanced by preprocessing the images to extract their edges, which exhibit robustness to seasonal and illumination variations.
1 code implementation • 10 Jan 2023 • Thomas Duboudin, Emmanuel Dellandréa, Corentin Abgrall, Gilles Hénaff, Liming Chen
Deep neural networks do not discriminate between spurious and causal patterns, and will only learn the most predictive ones while ignoring the others.
no code implementations • 17 Oct 2022 • Thomas Duboudin, Emmanuel Dellandréa, Corentin Abgrall, Gilles Hénaff, Liming Chen
Indeed, test-time adaptation methods usually have to rely on a limited representation because of the shortcut learning phenomenon: only a subset of the available predictive patterns is learned with standard training.
no code implementations • 15 Jun 2021 • Thomas Duboudin, Emmanuel Dellandréa, Corentin Abgrall, Gilles Hénaff, Liming Chen
Traditional deep learning algorithms often fail to generalize when they are tested outside of the domain of the training data.