Backdoor Attack with Sparse and Invisible Trigger
Preprint
Abstract
Deep neural networks (DNNs) are vulnerable to backdoor attacks, where the adversary manipulates a small portion of training data such that the victim model predicts normally on the benign samples but classifies the triggered samples as the target class. The backdoor attack is an emerging yet threatening training-phase threat, leading to serious risks in DNN-based applications. In this paper, we revisit the trigger patterns of existing backdoor attacks. We reveal that they are either visible or not sparse and therefore are not stealthy enough. More importantly, it is not feasible to simply combine existing methods to design an effective sparse and invisible backdoor attack. To address this problem, we formulate the trigger generation as a bi-level optimization problem with sparsity and invisibility constraints and propose an effective method to solve it. The proposed method is dubbed sparse and invisible backdoor attack (SIBA). We conduct extensive experiments on benchmark datasets under different settings, which verify the effectiveness of our attack and its resistance to existing backdoor defenses. The codes for reproducing main experiments are available at \url{https://github.com/YinghuaGao/SIBA}.Keywords
All Related Versions
This publication has 33 references indexed in Scilit:
- Backdoor Attacks Against Deep Learning Systems in the Physical World2021
- Bias-Based Universal Adversarial Patch Attack for Automatic Check-Out2020
- Sparse Adversarial Attack via Perturbation Factorization2020
- Automatic Neural Network Compression by Sparsity-Quantization Joint Learning: A Constrained Optimization-Based Approach2020
- Clean-Label Backdoor Attacks on Video Recognition Models2020
- Compressing Convolutional Neural Networks via Factorized Convolutional Filters2019
- BadNets: Evaluating Backdooring Attacks on Deep Neural NetworksIEEE Access, 2019
- Fine-Pruning: Defending Against Backdooring Attacks on Deep Neural Networks2018
- Deep Residual Learning for Image Recognition2016
- Data compression and harmonic analysisIEEE Transactions on Information Theory, 1998