Adversarially Robust Spiking Neural Networks with Sparse Connectivity

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Schmolli, Mathias, Baronig, Maximilian, Legenstein, Robert, Özdenizci, Ozan
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908374019342336
author Schmolli, Mathias
Baronig, Maximilian
Legenstein, Robert
Özdenizci, Ozan
author_facet Schmolli, Mathias
Baronig, Maximilian
Legenstein, Robert
Özdenizci, Ozan
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2505_15833
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Adversarially Robust Spiking Neural Networks with Sparse Connectivity
Schmolli, Mathias
Baronig, Maximilian
Legenstein, Robert
Özdenizci, Ozan
Neural and Evolutionary Computing
Cryptography and Security
Machine Learning
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.
title Adversarially Robust Spiking Neural Networks with Sparse Connectivity
topic Neural and Evolutionary Computing
Cryptography and Security
Machine Learning
url https://arxiv.org/abs/2505.15833