Adversarial Training for Face Recognition Systems using Contrastive Adversarial Learning and Triplet Loss Fine-tuning

TL;DR

This paper introduces a novel self-supervised adversarial training approach combining contrastive pre-training with triplet loss fine-tuning to enhance face recognition robustness efficiently.

Contribution

It presents a new method that improves adversarial robustness in face recognition with fewer fine-tuning epochs compared to traditional methods.

Findings

Comparable robustness with fewer epochs during fine-tuning

Contrastive adversarial training benefits from small amounts of labeled data

Method outperforms standard pre-trained models in adversarial settings

Abstract

Though much work has been done in the domain of improving the adversarial robustness of facial recognition systems, a surprisingly small percentage of it has focused on self-supervised approaches. In this work, we present an approach that combines Ad-versarial Pre-Training with Triplet Loss AdversarialFine-Tuning. We compare our methods with the pre-trained ResNet50 model that forms the backbone of FaceNet, finetuned on our CelebA dataset. Through comparing adversarial robustness achieved without adversarial training, with triplet loss adversarial training, and our contrastive pre-training combined with triplet loss adversarial fine-tuning, we find that our method achieves comparable results with far fewer epochs re-quired during fine-tuning. This seems promising, increasing the training time for fine-tuning should yield even better results. In addition to this, a modified semi-supervised experiment was conducted, which demonstrated the improvement of contrastive adversarial training with the introduction of small amounts of labels.