Haoyu Wang, Zhuo Huang, Xiaolong Wang, Bo Han, Zhiwei Lin, Tongliang Liu
MeGU is a new framework for machine unlearning that uses multi-modal large language models to selectively erase target data influence while preserving model utility.
The paper addresses the challenge of removing specific data from machine learning models without compromising their performance. Traditional methods either remove too much data, harming the model, or leave some unwanted data behind. The authors propose a new approach called Machine-Guided Unlearning (MeGU), which uses advanced language models to guide the process of unlearning. This method helps ensure that only the intended data is forgotten, while the model retains its usefulness and accuracy.
