ContinualFlow: Learning and Unlearning with Neural Flow Matching
This addresses the need for efficient unlearning in generative models, which is incremental as it builds on existing Flow Matching methods.
The paper tackles the problem of targeted unlearning in generative models by introducing ContinualFlow, a framework that uses energy-based reweighting to subtract undesired data regions without retraining or direct sample access, validated through experiments on 2D and image domains.
We introduce ContinualFlow, a principled framework for targeted unlearning in generative models via Flow Matching. Our method leverages an energy-based reweighting loss to softly subtract undesired regions of the data distribution without retraining from scratch or requiring direct access to the samples to be unlearned. Instead, it relies on energy-based proxies to guide the unlearning process. We prove that this induces gradients equivalent to Flow Matching toward a soft mass-subtracted target, and validate the framework through experiments on 2D and image domains, supported by interpretable visualizations and quantitative evaluations.