CorruptEncoder: Data Poisoning based Backdoor Attacks to Contrastive Learning. (arXiv:2211.08229v1 [cs.CR])

Contrastive learning (CL) pre-trains general-purpose encoders using an
unlabeled pre-training dataset, which consists of images (called single-modal
CL) or image-text pairs (called multi-modal CL). CL is vulnerable to data
poisoning based backdoor attacks (DPBAs), in which an attacker injects poisoned
inputs into the pre-training dataset so the encoder is backdoored. However,
existing DPBAs achieve limited effectiveness. In this work, we propose new
DPBAs called CorruptEncoder to CL. Our experiments show that CorruptEncoder
substantially outperforms existing DPBAs for both single-modal and …

attacks backdoor data data poisoning poisoning