TMI! Finetuned Models Leak Private Information from their Pretraining Data

arXiv:2306.01181v2 Announce Type: replace
Abstract: Transfer learning has become an increasingly popular technique in machine learning as a way to leverage a pretrained model trained for one task to assist with building a finetuned model for a related task. This paradigm has been especially popular for $\textit{privacy}$ in machine learning, where the pretrained model is considered public, and only the data for finetuning is considered sensitive. However, there are reasons to believe that the data used for pretraining is still …

arxiv cs.cr cs.lg data information leak pretraining type