Robust and Private Learning of Halfspaces

Badih Ghazi · Ravi Kumar · Pasin Manurangsi · Thao Nguyen

Keywords: [ Ethics and Safety ] [ Privacy-preserving Statistics and Machine Learning ]

[ Abstract ]
Wed 14 Apr 6 a.m. PDT — 8 a.m. PDT
Oral presentation: Bandits, Reinforcement Learning / Learning Theory / Sparse Methods
Wed 14 Apr 9:15 a.m. PDT — 10:15 a.m. PDT

Abstract: In this work, we study the trade-off between differential privacy and adversarial robustness under $L_2$-perturbations in the context of learning halfspaces. We prove nearly tight bounds on the sample complexity of robust private learning of halfspaces for a large regime of parameters. A highlight of our results is that robust and private learning is harder than robust or private learning alone. We complement our theoretical analysis with experimental results on the MNIST and USPS datasets, for a learning algorithm that is both differentially private and adversarially robust.

Chat is not available.