Deep semi-parametric models allow for the instant deletion of training data from a model without retraining or parameter updates.
March 25, 2026
Original Paper
Designing to Forget: Deep Semi-parametric Models for Unlearning
arXiv · 2603.22870
The Takeaway
This architecture enables 'designing to forget' by aggregating training sample info in a fusion module, allowing for explicit test-time deletion. It achieves 10x faster unlearning compared to parametric methods while maintaining competitive performance on ImageNet.
From the abstract
Recent advances in machine unlearning have focused on developing algorithms to remove specific training samples from a trained model. In contrast, we observe that not all models are equally easy to unlearn. Hence, we introduce a family of deep semi-parametric models (SPMs) that exhibit non-parametric behavior during unlearning. SPMs use a fusion module that aggregates information from each training sample, enabling explicit test-time deletion of selected samples without altering model parameters