Controllable Unlearning for Image-to-Image Generative Models via $\varepsilon$-Constrained Optimization
This work addresses privacy and bias concerns in generative models for users needing flexible unlearning, but it is incremental as it builds on existing unlearning methods by adding controllability.
The paper tackles the problem of machine unlearning in Image-to-Image generative models by proposing a controllable framework that uses a control coefficient to manage the trade-off between complete unlearning and model utility, achieving Pareto optimal solutions across benchmark datasets and models.
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.