Adversarial Machine Learning And Speech Emotion Recognition: Utilizing Generative Adversarial Networks For Robustness

TL;DR

This paper investigates the vulnerability of speech emotion recognition systems to adversarial attacks and proposes using GANs to improve their robustness through novel defense strategies.

Contribution

It introduces the first black-box adversarial attack on SER systems and explores GAN-based defenses to enhance their robustness against adversarial examples.

Findings

Adversarial examples can successfully attack SER systems.

GAN-based defenses improve robustness of SER systems.

The study opens new avenues for adversarial robustness in speech emotion recognition.

Abstract

Deep learning has undoubtedly offered tremendous improvements in the performance of state-of-the-art speech emotion recognition (SER) systems. However, recent research on adversarial examples poses enormous challenges on the robustness of SER systems by showing the susceptibility of deep neural networks to adversarial examples as they rely only on small and imperceptible perturbations. In this study, we evaluate how adversarial examples can be used to attack SER systems and propose the first black-box adversarial attack on SER systems. We also explore potential defenses including adversarial training and generative adversarial network (GAN) to enhance robustness. Experimental evaluations suggest various interesting aspects of the effective utilization of adversarial examples useful for achieving robustness for SER systems opening up opportunities for researchers to further innovate in this space.