Adversarially Robust Spiking Neural Networks with Sparse Connectivity

TL;DR

This paper presents a method to convert robust artificial neural networks into sparse, energy-efficient spiking neural networks that maintain high adversarial robustness, significantly reducing memory and energy consumption.

Contribution

The authors introduce a novel conversion algorithm that creates sparse, adversarially robust SNNs from pretrained ANNs, combining energy efficiency with robustness.

Findings

Achieved up to 100x reduction in stored weights.

Realized approximately 8.6x increase in energy efficiency.

Maintained high adversarial robustness and performance.

Abstract

Deployment of deep neural networks in resource-constrained embedded systems requires innovative algorithmic solutions to facilitate their energy and memory efficiency. To further ensure the reliability of these systems against malicious actors, recent works have extensively studied adversarial robustness of existing architectures. Our work focuses on the intersection of adversarial robustness, memory- and energy-efficiency in neural networks. We introduce a neural network conversion algorithm designed to produce sparse and adversarially robust spiking neural networks (SNNs) by leveraging the sparse connectivity and weights from a robustly pretrained artificial neural network (ANN). Our approach combines the energy-efficient architecture of SNNs with a novel conversion algorithm, leading to state-of-the-art performance with enhanced energy and memory efficiency through sparse connectivity and activations. Our models are shown to achieve up to 100x reduction in the number of weights to be stored in memory, with an estimated 8.6x increase in energy efficiency compared to dense SNNs, while maintaining high performance and robustness against adversarial threats.