Summary of Where Did Your Model Learn That? Label-free Influence For Self-supervised Learning, by Nidhin Harilal et al.
Where Did Your Model Learn That? Label-free Influence for Self-supervised Learning
by Nidhin Harilal, Amit Kiran Rege, Reza Akbarian Bafghi, Maziar Raissi, Claire Monteleoni
First submitted to arxiv on: 22 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computer Vision and Pattern Recognition (cs.CV)
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 The proposed Influence-SSL approach enables label-free analysis of self-supervised learning (SSL) models, allowing for the identification of training examples that significantly influence model predictions. By harnessing the stability of learned representations against data augmentations, Influence-SSL provides a novel method for defining influence functions tailored to SSL settings. Theoretical foundations and empirical evidence demonstrate the utility of Influence-SSL in analyzing pre-trained SSL models, revealing differences in how SSL models respond to influential data compared to supervised models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Influence-SSL is a new way to understand how self-supervised learning works without needing labels. It helps identify which training examples are most important for making predictions. This method uses the stability of learned representations against small changes to see which examples matter most. By doing so, it provides insights into pre-trained SSL models and shows that they respond differently to influential data than supervised models do. |
Keywords
» Artificial intelligence » Self supervised » Supervised
