pith. sign in

arxiv: 1905.05897 · v2 · pith:Q3FUIXNZnew · submitted 2019-05-15 · 📊 stat.ML · cs.CR· cs.LG

Transferable Clean-Label Poisoning Attacks on Deep Neural Nets

classification 📊 stat.ML cs.CRcs.LG
keywords poisoningattackattacksdatapoisontrainingtransferableachieve
0
0 comments X
read the original abstract

Clean-label poisoning attacks inject innocuous looking (and "correctly" labeled) poison images into training data, causing a model to misclassify a targeted image after being trained on this data. We consider transferable poisoning attacks that succeed without access to the victim network's outputs, architecture, or (in some cases) training data. To achieve this, we propose a new "polytope attack" in which poison images are designed to surround the targeted image in feature space. We also demonstrate that using Dropout during poison creation helps to enhance transferability of this attack. We achieve transferable attack success rates of over 50% while poisoning only 1% of the training set.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.