Machine Unlearning under Retain-Forget Entanglement

TL;DR

This paper introduces a two-phase optimization framework for machine unlearning that effectively handles entanglements between forget and retain samples, ensuring accurate removal and retention of information.

Contribution

It presents a novel two-phase method combining augmented Lagrangian and Wasserstein-regularized gradient projection to improve unlearning performance amidst sample entanglements.

Findings

Outperforms existing methods in accuracy retention and removal fidelity.

Effective across multiple datasets and neural architectures.

Achieves reliable unlearning with minimal performance degradation.

Abstract

Forgetting a subset in machine unlearning is rarely an isolated task. Often, retained samples that are closely related to the forget set can be unintentionally affected, particularly when they share correlated features from pretraining or exhibit strong semantic similarities. To address this challenge, we propose a novel two-phase optimization framework specifically designed to handle such retai-forget entanglements. In the first phase, an augmented Lagrangian method increases the loss on the forget set while preserving accuracy on less-related retained samples. The second phase applies a gradient projection step, regularized by the Wasserstein-2 distance, to mitigate performance degradation on semantically related retained samples without compromising the unlearning objective. We validate our approach through comprehensive experiments on multiple unlearning tasks, standard benchmark datasets, and diverse neural architectures, demonstrating that it achieves effective and reliable unlearning while outperforming existing baselines in both accuracy retention and removal fidelity.