NegMerge: Sign-Consensual Weight Merging for Machine Unlearning

TL;DR

NegMerge introduces a sign-consensual weight merging technique for machine unlearning that leverages multiple fine-tuned models to improve effectiveness and efficiency across various datasets and architectures.

Contribution

The paper proposes a novel sign-consensual merging method that aggregates multiple fine-tuned models' task vectors for more robust unlearning, reducing hyperparameter sensitivity.

Findings

Outperforms state-of-the-art unlearning methods on multiple datasets.

Requires similar or fewer computational resources than existing approaches.

Effective across diverse architectures and tasks.

Abstract

Machine unlearning aims to selectively remove specific knowledge from a trained model. Existing approaches, such as Task Arithmetic, fine-tune the model on the forget set to create a task vector (i.e., a direction in weight space) for subtraction from the original model's weight. However, their effectiveness is highly sensitive to hyperparameter selection, requiring extensive validation to identify the optimal vector from many fine-tuned candidates. In this paper, we propose a novel method that utilizes all fine-tuned models trained with varying hyperparameters instead of a single selection. Specifically, we aggregate the computed task vectors by retaining only the elements with consistent shared signs. The merged task vector is then negated to induce unlearning on the original model. Evaluations on zero-shot and standard image recognition tasks across twelve datasets and four backbone architectures show that our approach outperforms state-of-the-art methods while requiring similar or fewer computational resources. Code is available at https://github.com/naver-ai/negmerge.