A_Self_supervised_Approach_for_Adversarial_Robustness.md

@inproceedings{Naseer_2020_CVPR,
author = {Naseer, Muzammal and Khan, Salman and Hayat, Munawar and Khan, Fahad Shahbaz and Porikli, Fatih},
booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {jun},
title = {{A Self-supervised Approach for Adversarial Robustness}},
year = {2020}
}

Motivation

• Adversarial training that enhances robustness by modifying target model's parameters lacks generalizability of cross-task protection.
• Different input processing based defenses fall short in the face of continuously evolving attacks.
• Our defense aims to combine the benefits of adversarial training and input processing methods in a single frame- work that is computationally efficient, generalizable across different tasks and retains the clean image accuracy.
• Combine the pre-processing and adversarial training.
• build a robust denosier

Contribution

• A self-supervised way to generate adversarial perturbations, which is proved to be transferable.
• using the adversarial training scheme to train the robust purifier.