Adversarial Attacks on Spiking Convolutional Neural Networks for Event-based Vision

By Julian B., et al

Read the original document by opening this link in a new tab.

Table of Contents

Introduction
What is event-based sensing?
Adversarial attacks on discrete data
Attack strategies
SparseFool on discrete data
Adversarial patches
Datasets
Network models

Summary

This paper explores adversarial attacks on spiking convolutional neural networks for event-based vision. The study focuses on the vulnerability of spiking neural networks to adversarial attacks and demonstrates the adaptation of attack algorithms to event-based visual data. The authors show smaller perturbation magnitudes and higher success rates compared to existing algorithms. They also test the effectiveness of these perturbations on neuromorphic hardware, emphasizing the importance of ensuring the security and reliability of neuromorphic vision devices in various contexts.