Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Mudjacking: Patching Backdoor Vulnerabilities in Foundation Models

About

Foundation model has become the backbone of the AI ecosystem. In particular, a foundation model can be used as a general-purpose feature extractor to build various downstream classifiers. However, foundation models are vulnerable to backdoor attacks and a backdoored foundation model is a single-point-of-failure of the AI ecosystem, e.g., multiple downstream classifiers inherit the backdoor vulnerabilities simultaneously. In this work, we propose Mudjacking, the first method to patch foundation models to remove backdoors. Specifically, given a misclassified trigger-embedded input detected after a backdoored foundation model is deployed, Mudjacking adjusts the parameters of the foundation model to remove the backdoor. We formulate patching a foundation model as an optimization problem and propose a gradient descent based method to solve it. We evaluate Mudjacking on both vision and language foundation models, eleven benchmark datasets, five existing backdoor attacks, and thirteen adaptive backdoor attacks. Our results show that Mudjacking can remove backdoor from a foundation model while maintaining its utility.

Hongbin Liu, Michael K. Reiter, Neil Zhenqiang Gong• 2024

Related benchmarks

TaskDatasetResultRank
Code GenerationHumanEval (test)
Pass@161
701
Sentiment AnalysisSST-2 (test)
Accuracy93
162
Sentiment AnalysisSST2
ASR97
131
Linguistic AcceptabilityCOLA
Accuracy (CoLA)81
108
Text ClassificationSST-2
CACC93
80
Topic ClassificationAGNews
ASR0.02
78
Jailbreak DefenseLlama model jailbreak evaluation prompts
ASR5
60
Safety Defense EvaluationMATH
ASR29
60
Safety Defense EvaluationMedicine
ASR48
60
Sentiment AnalysisSST2 UJB Attack (test)
ASR41
14
Showing 10 of 19 rows

Other info

Follow for update