Summary of Is Retain Set All You Need in Machine Unlearning? Restoring Performance Of Unlearned Models with Out-of-distribution Images, by Jacopo Bonato et al.
Is Retain Set All You Need in Machine Unlearning? Restoring Performance of Unlearned Models with Out-Of-Distribution Images
by Jacopo Bonato, Marco Cotogni, Luigi Sabetta
First submitted to arxiv on: 19 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary A novel approximate unlearning method called Selective-distillation for Class and Architecture-agnostic unleaRning (SCAR) is introduced in this paper. SCAR efficiently eliminates specific information while preserving the model’s test accuracy without using a retain set, which is a key component in state-of-the-art approximate unlearning algorithms. The approach utilizes a modified Mahalanobis distance to guide the unlearning of feature vectors, and a distillation-trick mechanism that distills knowledge from the original model into the unlearning model with out-of-distribution images. A self-forget version of SCAR is also proposed, which unlearns without having access to the forget set. Experimental results on three public datasets demonstrate the effectiveness of SCAR, achieving performance higher than methods operating without a retain set and comparable to state-of-the-art methods that rely on it. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary SCAR is a new way to remove specific information from models while keeping their overall accuracy. This helps prevent models from using unwanted knowledge. The method uses a special distance measure to guide the removal process, and a trick to copy important information from one model to another. It also has a version that can remove information without needing access to the original data. Tests on three big datasets show that SCAR works well. |
Keywords
» Artificial intelligence » Distillation
