Skip to main navigation Skip to search Skip to main content

IBD-PSC: Input-level Backdoor Detection via Parameter-oriented Scaling Consistency

  • Linshan Hou
  • , Ruili Feng
  • , Zhongyun Hua*
  • , Wei Luo
  • , Leo Yu Zhang
  • , Yiming Li*
  • *Corresponding author for this work
  • School of Computer Science and Technology, Harbin Institute of Technology
  • Alibaba Group Holding Ltd.
  • University of Science and Technology of China
  • Deakin University
  • Griffith University Queensland
  • Nanyang Technological University

Research output: Contribution to journalConference articlepeer-review

Abstract

Deep neural networks (DNNs) are vulnerable to backdoor attacks, where adversaries can maliciously trigger model misclassifications by implanting a hidden backdoor during model training. This paper proposes a simple yet effective input-level backdoor detection (dubbed IBD-PSC) as a 'firewall' to filter out malicious testing images. Our method is motivated by an intriguing phenomenon, i.e., parameter-oriented scaling consistency (PSC), where the prediction confidences of poisoned samples are significantly more consistent than those of benign ones when amplifying model parameters. In particular, we provide theoretical analysis to safeguard the foundations of the PSC phenomenon. We also design an adaptive method to select BN layers to scale up for effective detection. Extensive experiments are conducted on benchmark datasets, verifying the effectiveness and efficiency of our IBD-PSC method and its resistance to adaptive attacks. Codes are available at BackdoorBox.

Original languageEnglish
Pages (from-to)18992-19022
Number of pages31
JournalProceedings of Machine Learning Research
Volume235
StatePublished - 2024
Externally publishedYes
Event41st International Conference on Machine Learning, ICML 2024 - Vienna, Austria
Duration: 21 Jul 202427 Jul 2024

Fingerprint

Dive into the research topics of 'IBD-PSC: Input-level Backdoor Detection via Parameter-oriented Scaling Consistency'. Together they form a unique fingerprint.

Cite this