MUSO: Achieving Exact Machine Unlearning in Over-Parameterized Regimes
This addresses the need for precise data removal in machine learning models, particularly for privacy and compliance, with incremental improvements in method efficiency.
The paper tackles the problem of achieving exact machine unlearning in over-parameterized models, demonstrating theoretically that over-parameterized linear models can achieve exact unlearning through data relabeling and proposing an algorithm that outperforms state-of-the-art methods in numerical experiments.
Machine unlearning (MU) is to make a well-trained model behave as if it had never been trained on specific data. In today's over-parameterized models, dominated by neural networks, a common approach is to manually relabel data and fine-tune the well-trained model. It can approximate the MU model in the output space, but the question remains whether it can achieve exact MU, i.e., in the parameter space. We answer this question by employing random feature techniques to construct an analytical framework. Under the premise of model optimization via stochastic gradient descent, we theoretically demonstrated that over-parameterized linear models can achieve exact MU through relabeling specific data. We also extend this work to real-world nonlinear networks and propose an alternating optimization algorithm that unifies the tasks of unlearning and relabeling. The algorithm's effectiveness, confirmed through numerical experiments, highlights its superior performance in unlearning across various scenarios compared to current state-of-the-art methods, particularly excelling over similar relabeling-based MU approaches.