When Machine Learning Models Leak: An Exploration of Synthetic Training Data

arXiv:2310.08775v2 Announce Type: replace
Abstract: We investigate an attack on a machine learning model that predicts whether a person or household will relocate in the next two years, i.e., a propensity-to-move classifier. The attack assumes that the attacker can query the model to obtain predictions and that the marginal distribution of the data on which the model was trained is publicly available. The attack also assumes that the attacker has obtained the values of non-sensitive attributes for a certain number …

abstract arxiv classifier cs.lg data exploration leak machine machine learning machine learning model machine learning models next person predictions query synthetic training training data type will