English

Neural Architectural Backdoors

Cryptography and Security 2022-11-08 v2 Machine Learning

Abstract

This paper asks the intriguing question: is it possible to exploit neural architecture search (NAS) as a new attack vector to launch previously improbable attacks? Specifically, we present EVAS, a new attack that leverages NAS to find neural architectures with inherent backdoors and exploits such vulnerability using input-aware triggers. Compared with existing attacks, EVAS demonstrates many interesting properties: (i) it does not require polluting training data or perturbing model parameters; (ii) it is agnostic to downstream fine-tuning or even re-training from scratch; (iii) it naturally evades defenses that rely on inspecting model parameters or training data. With extensive evaluation on benchmark datasets, we show that EVAS features high evasiveness, transferability, and robustness, thereby expanding the adversary's design spectrum. We further characterize the mechanisms underlying EVAS, which are possibly explainable by architecture-level ``shortcuts'' that recognize trigger patterns. This work raises concerns about the current practice of NAS and points to potential directions to develop effective countermeasures.

Keywords

Cite

@article{arxiv.2210.12179,
  title  = {Neural Architectural Backdoors},
  author = {Ren Pang and Changjiang Li and Zhaohan Xi and Shouling Ji and Ting Wang},
  journal= {arXiv preprint arXiv:2210.12179},
  year   = {2022}
}
R2 v1 2026-06-28T04:12:42.099Z