BadEncoder: Backdoor Attacks to Pre-trained Encoders in Self-Supervised Learning

Jinyuan Jia, Yupei Liu, Neil Zhenqiang Gong

Research output: Chapter in Book/Report/Conference proceedingConference contribution

28 Scopus citations


Self-supervised learning in computer vision aims to pre-train an image encoder using a large amount of unlabeled images or (image, text) pairs. The pre-trained image encoder can then be used as a feature extractor to build downstream classifiers for many downstream tasks with a small amount of or no labeled training data. In this work, we propose BadEncoder, the first backdoor attack to self-supervised learning. In particular, our BadEncoder injects backdoors into a pre-trained image encoder such that the downstream classifiers built based on the backdoored image encoder for different downstream tasks simultaneously inherit the backdoor behavior. We formulate our BadEncoder as an optimization problem and we propose a gradient descent based method to solve it, which produces a backdoored image encoder from a clean one. Our extensive empirical evaluation results on multiple datasets show that our BadEncoder achieves high attack success rates while preserving the accuracy of the downstream classifiers. We also show the effectiveness of BadEncoder using two publicly available, real-world image encoders, i.e., Google's image encoder pre-trained on ImageNet and OpenAI's Contrastive Language-Image Pre-training (CLIP) image encoder pre-trained on 400 million (image, text) pairs collected from the Internet. Moreover, we consider defenses including Neural Cleanse and MNTD (empirical defenses) as well as PatchGuard (a provable defense). Our results show that these defenses are insufficient to defend against BadEncoder, highlighting the needs for new defenses against our BadEncoder. Our code is publicly available at:

Original languageEnglish (US)
Title of host publicationProceedings - 43rd IEEE Symposium on Security and Privacy, SP 2022
PublisherInstitute of Electrical and Electronics Engineers Inc.
Number of pages17
ISBN (Electronic)9781665413169
StatePublished - 2022
Event43rd IEEE Symposium on Security and Privacy, SP 2022 - San Francisco, United States
Duration: May 23 2022May 26 2022

Publication series

NameProceedings - IEEE Symposium on Security and Privacy
ISSN (Print)1081-6011


Conference43rd IEEE Symposium on Security and Privacy, SP 2022
Country/TerritoryUnited States
CitySan Francisco

All Science Journal Classification (ASJC) codes

  • Safety, Risk, Reliability and Quality
  • Software
  • Computer Networks and Communications

Cite this