Toward Adversarial Robustness by Diversity in an Ensemble of Specialized Deep Neural Networks

TL;DR

This paper demonstrates that diversity in an ensemble of specialized CNNs enhances adversarial detection and robustness, reducing the success of both black-box and white-box attacks on datasets like MNIST and CIFAR-10.

Contribution

It introduces a diverse ensemble of specialized CNNs with a voting mechanism, showing improved adversarial detection and robustness over traditional ensembles.

Findings

Effective detection of black-box adversarial examples.

Significant reduction in white-box attack success rates.

Small increase in misclassification of clean samples.

Abstract

We aim at demonstrating the influence of diversity in the ensemble of CNNs on the detection of black-box adversarial instances and hardening the generation of white-box adversarial attacks. To this end, we propose an ensemble of diverse specialized CNNs along with a simple voting mechanism. The diversity in this ensemble creates a gap between the predictive confidences of adversaries and those of clean samples, making adversaries detectable. We then analyze how diversity in such an ensemble of specialists may mitigate the risk of the black-box and white-box adversarial examples. Using MNIST and CIFAR-10, we empirically verify the ability of our ensemble to detect a large portion of well-known black-box adversarial examples, which leads to a significant reduction in the risk rate of adversaries, at the expense of a small increase in the risk rate of clean samples. Moreover, we show that the success rate of generating white-box attacks by our ensemble is remarkably decreased compared to a vanilla CNN and an ensemble of vanilla CNNs, highlighting the beneficial role of diversity in the ensemble for developing more robust models.