Architectural Backdoors in Neural Networks
Authors:
- 1 June 2023
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 24595-24604
Abstract
Machine learning is vulnerable to adversarial manipulation. Previous literature demonstrated that at the training stage attackers can manipulate data [14] and data sampling procedures [29] to control model behaviour. A common attack goal is to plant backdoors i.e. force the victim model to learn to recognise a trigger known only by the adversary. In this paper, we introduce a new class of backdoor attacks that hide inside model architectures i.e. in the inductive bias of the functions used to train. These backdoors are simple to implement, for instance by publishing open-source code for a backdoored model architecture that others will reuse unknowingly. We demonstrate that model architectural backdoors represent a real threat and, unlike other approaches, can survive a complete re-training from scratch. We formalise the main construction principles behind architectural backdoors, such as a connection between the input and the output, and describe some possible protections against them. We evaluate our attacks on computer vision benchmarks of different scales and demonstrate the underlying vulnerability is pervasive in a variety of common training settings.Keywords
All Related Versions
This publication has 13 references indexed in Scilit:
- Februus: Input Purification Defense Against Trojan Attacks on Deep Neural Network Systems2020
- Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks2019
- Wild patterns: Ten years after the rise of adversarial machine learningPattern Recognition, 2018
- Methods for interpreting and understanding deep neural networksDigital Signal Processing, 2018
- Towards Evaluating the Robustness of Neural Networks2017
- Membership Inference Attacks Against Machine Learning Models2017
- Practical Black-Box Attacks against Machine Learning2017
- DEX: Deep EXpectation of Apparent Age from a Single Image2015
- Traffic sign recognition — How far are we from the solution?2013
- Man vs. computer: Benchmarking machine learning algorithms for traffic sign recognitionNeural Networks, 2012