Controllable Unlearning for Image-to-Image Generative Models via $\varepsilon$-Constrained Optimization

TL;DR

This paper introduces a controllable unlearning framework for image-to-image generative models that balances unlearning of specific data with model utility using an $ ext{ε}$-constrained optimization approach, ensuring Pareto optimality.

Contribution

It formulates the unlearning problem as an $ ext{ε}$-constrained optimization, providing a controllable trade-off mechanism and theoretical guarantees for solutions.

Findings

Effective unlearning demonstrated on benchmark datasets

Framework achieves Pareto optimal solutions

Convergence rate analyzed under different control functions

Abstract

While generative models have made significant advancements in recent years, they also raise concerns such as privacy breaches and biases. Machine unlearning has emerged as a viable solution, aiming to remove specific training data, e.g., containing private information and bias, from models. In this paper, we study the machine unlearning problem in Image-to-Image (I2I) generative models. Previous studies mainly treat it as a single objective optimization problem, offering a solitary solution, thereby neglecting the varied user expectations towards the trade-off between complete unlearning and model utility. To address this issue, we propose a controllable unlearning framework that uses a control coefficient $\varepsilon$ to control the trade-off. We reformulate the I2I generative model unlearning problem into a $\varepsilon$-constrained optimization problem and solve it with a gradient-based method to find optimal solutions for unlearning boundaries. These boundaries define the valid range for the control coefficient. Within this range, every yielded solution is theoretically guaranteed with Pareto optimality. We also analyze the convergence rate of our framework under various control functions. Extensive experiments on two benchmark datasets across three mainstream I2I models demonstrate the effectiveness of our controllable unlearning framework.