# Stochastic sparse adversarial attacks

arXiv: https://arxiv.org/abs/2011.12423 · HTML: https://mlsanity.com/paper/2011.12423/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 24 Nov 2020.

## Built on by (most cited first, 4 of 4)

- [AttackBench: Evaluating Gradient-based Attacks for Adversarial Examples](https://mlsanity.com/paper/2404.19460/) (2024)
- [σ-zero: Gradient-based Optimization of ℓ0-norm Adversarial Examples](https://mlsanity.com/paper/2402.01879/) (2024)
- [Improving Robustness of Deep Reinforcement Learning Agents: Environment Attack based on the Critic Network](https://mlsanity.com/paper/2104.03154/) (2021)
- [Riemannian data-dependent randomized smoothing for neural networks certification](https://mlsanity.com/paper/2206.10235/) (2022)

## Abstract (arXiv)

This paper introduces stochastic sparse adversarial attacks (SSAA), standing as simple, fast and purely noise-based targeted and untargeted attacks of neural network classifiers (NNC). SSAA offer new examples of sparse (or L0) attacks for which only few methods have been proposed previously. These attacks are devised by exploiting a small-time expansion idea widely used for Markov processes. Experiments on small and large datasets (CIFAR-10 and ImageNet) illustrate several advantages of SSAA in comparison with the-state-of-the-art methods. For instance, in the untargeted case, our method called Voting Folded Gaussian Attack (VFGA) scales efficiently to ImageNet and achieves a significantly lower L0 score than SparseFool (up to frac25) while being faster. Moreover, VFGA achieves better L0 scores on ImageNet than Sparse-RS when both attacks are fully successful on a large number of samples.