Addressing Neural Network Robustness with Mixup and Targeted Labeling Adversarial Training

Alfred Laugros (1, 2) , Alice Caplier (1) , Matthieu Ospici (2)

1 GIPSA-ACTIV - GIPSA - Apprentissage, Classification, Traitement d'Images et de Vidéos
2 Atos

Despite their performance, Artificial Neural Networks are not reliable enough for most of industrial applications. They are sensitive to noises, rotations, blurs and adversarial examples. There is a need to build defenses that protect against a wide range of perturbations, covering the most traditional common corruptions and adversarial examples. We propose a new data augmentation strategy called M-TLAT and designed to address robustness in a broad sense. Our approach combines the Mixup augmentation and a new adversarial training algorithm called Targeted Labeling Adversarial Training (TLAT). The idea of TLAT is to interpolate the target labels of adversarial examples with the ground-truth labels. We show that M-TLAT can increase the robustness of image classifiers towards nineteen common corruptions and five adversarial attacks , without reducing the accuracy on clean samples.

Mots clés: Neural Network; Robustness; Common Corruptions; Adversarial Training; Mixup

Domaines: Intelligence artificielle [cs.AI]; Vision par ordinateur et reconnaissance de formes [cs.CV]

Alfred Laugros, Alice Caplier, Matthieu Ospici. Addressing Neural Network Robustness with Mixup and Targeted Labeling Adversarial Training. ECCV 2020 - 16th European Conference on Computer Vision, Aug 2020, Glasgow, United Kingdom. ⟨10.1007/978-3-030-68238-5_14⟩. ⟨hal-02925252⟩