Langevin Unlearning: A New Perspective of Noisy Gradient Descent for Machine Unlearning

Eli Chien, Haoyu Peter Wang, Ziang Chen, Pan Li

Keywords: Machine unlearning, privacy, Langevin Monte Carlo, Langevin dynamic, gradient descent, differential privacy

TL;DR: We propose Langevin unlearning which not only connect DP with unlearning, but also provides unlearning privacy guarantees for PNGD with Langevin dynamic analysis.

Abstract: Machine unlearning has raised significant interest with the adoption of laws ensuring the ``right to be forgotten''. Researchers have provided a probabilistic notion of approximate unlearning under a similar definition of Differential Privacy (DP), where privacy is defined as statistical indistinguishability to retraining from scratch. We propose Langevin unlearning, an unlearning framework based on noisy gradient descent with privacy guarantees for approximate unlearning problems. Langevin unlearning unifies the DP learning process and the privacy-certified unlearning process with many algorithmic benefits. These include approximate certified unlearning for non-convex problems, complexity saving compared to retraining, sequential and batch unlearning for multiple unlearning requests.

Primary Area: Privacy

Submission Number: 7029