Remember What You Want to Forget: Algorithms for Machine Unlearning

TL;DR

This paper introduces algorithms for machine unlearning, enabling models to forget specific data points efficiently while maintaining accuracy, and highlights a fundamental difference from differential privacy in unlearning capabilities.

Contribution

The paper provides the first rigorous analysis of generalization in machine unlearning and presents an unlearning algorithm with superior sample deletion guarantees for convex losses.

Findings

Unlearning algorithm can remove up to O(n/d^{1/4}) samples.

Differential privacy guarantees only O(n/d^{1/2}) sample deletion.

Demonstrates a separation between differential privacy and machine unlearning.

Abstract

We study the problem of unlearning datapoints from a learnt model. The learner first receives a dataset $S$ drawn i.i.d. from an unknown distribution, and outputs a model $\widehat{w}$ that performs well on unseen samples from the same distribution. However, at some point in the future, any training datapoint $z \in S$ can request to be unlearned, thus prompting the learner to modify its output model while still ensuring the same accuracy guarantees. We initiate a rigorous study of generalization in machine unlearning, where the goal is to perform well on previously unseen datapoints. Our focus is on both computational and storage complexity. For the setting of convex losses, we provide an unlearning algorithm that can unlearn up to $O(n/d^{1/4})$ samples, where $d$ is the problem dimension. In comparison, in general, differentially private learning (which implies unlearning) only guarantees deletion of $O(n/d^{1/2})$ samples. This demonstrates a novel separation between differential privacy and machine unlearning.