Langevin Unlearning: A New Perspective of Noisy Gradient Descent for Machine Unlearning
This addresses the need for efficient and privacy-preserving unlearning in response to legal requirements like the 'right to be forgotten', though it appears incremental as it builds on existing DP-based approaches.
The paper tackles the problem of machine unlearning by proposing Langevin unlearning, a framework based on noisy gradient descent that provides privacy guarantees similar to Differential Privacy, resulting in benefits such as approximate certified unlearning for non-convex problems and complexity savings compared to retraining.
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.