R-SNN: An Analysis and Design Methodology for Robustifying Spiking Neural Networks against Adversarial Attacks through Noise Filters for Dynamic Vision Sensors

TL;DR

This paper introduces R-SNN, a methodology that enhances the robustness of Spiking Neural Networks against adversarial attacks on Dynamic Vision Sensors by employing noise filtering techniques, significantly improving accuracy under attack.

Contribution

The paper is the first to generate adversarial attacks on DVS signals and to apply noise filters for defending SNNs against such attacks.

Findings

Noise filters prevent SNNs from being fooled by adversarial attacks.

SNNs achieve over 90% accuracy on DVS-Gesture and NMNIST datasets under attack.

R-SNN improves robustness of SNNs in event-based vision systems.

Abstract

Spiking Neural Networks (SNNs) aim at providing energy-efficient learning capabilities when implemented on neuromorphic chips with event-based Dynamic Vision Sensors (DVS). This paper studies the robustness of SNNs against adversarial attacks on such DVS-based systems, and proposes R-SNN, a novel methodology for robustifying SNNs through efficient DVS-noise filtering. We are the first to generate adversarial attacks on DVS signals (i.e., frames of events in the spatio-temporal domain) and to apply noise filters for DVS sensors in the quest for defending against adversarial attacks. Our results show that the noise filters effectively prevent the SNNs from being fooled. The SNNs in our experiments provide more than 90% accuracy on the DVS-Gesture and NMNIST datasets under different adversarial threat models.