KGA: A General Machine Unlearning Framework Based on Knowledge Gap Alignment

TL;DR

This paper introduces KGA, a novel machine unlearning framework for NLP that maintains distribution differences to effectively forget specific training data, addressing privacy concerns in text-based models.

Contribution

KGA is the first general unlearning framework for NLP that relaxes distribution assumptions and applies across multiple NLP tasks with new evaluation metrics.

Findings

KGA outperforms baseline methods on large-scale NLP datasets.

Extensive experiments validate the effectiveness of KGA in various NLP tasks.

New evaluation metrics for unlearning in NLP are proposed.

Abstract

Recent legislation of the "right to be forgotten" has led to the interest in machine unlearning, where the learned models are endowed with the function to forget information about specific training instances as if they have never existed in the training set. Previous work mainly focuses on computer vision scenarios and largely ignores the essentials of unlearning in NLP field, where text data contains more explicit and sensitive personal information than images. In this paper, we propose a general unlearning framework called KGA to induce forgetfulness. Different from previous work that tries to recover gradients or forces models to perform close to one specific distribution, KGA maintains distribution differences (i.e., knowledge gap). This relaxes the distribution assumption. Furthermore, we first apply the unlearning method to various NLP tasks (i.e., classification, translation, response generation) and propose several unlearning evaluation metrics with pertinence. Experiments on large-scale datasets show that KGA yields comprehensive improvements over baselines, where extensive analyses further validate the effectiveness of KGA and provide insight into unlearning for NLP tasks.