BrainWash [1] is a recently proposed data poisoning attack designed to induce forgetting in task-incremental learners. This repository extends the BrainWash codebase with the implementation of PROACT, a defense strategy to counter the attack.
For each continual learning method, the defense logic is implemented in the defend_step function. The ewc_pipeline.sh script shows how to execute PROACT within the EWC pipeline.
[1] Abbasi, A., Nooralinejad, P., Pirsiavash, H., & Kolouri, S. (2024). Brainwash: A poisoning attack to forget in continual learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 24057-24067).