Summary of Exmap: Leveraging Explainability Heatmaps For Unsupervised Group Robustness to Spurious Correlations, by Rwiddhi Chakraborty et al.
ExMap: Leveraging Explainability Heatmaps for Unsupervised Group Robustness to Spurious Correlations
by Rwiddhi Chakraborty, Adrian Sletten, Michael Kampffmeyer
First submitted to arxiv on: 20 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: 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 ExMap, an unsupervised two-stage mechanism, is introduced to enhance group robustness in traditional classifiers. By utilizing clustering modules to infer pseudo-labels from a model’s explainability heatmaps, ExMap replaces actual labels during training. This approach bridges the performance gap with supervised counterparts and outperforms existing partially supervised and unsupervised methods. The efficacy of ExMap is empirically validated through studies demonstrating its potential in tackling multiple shortcut mitigation issues. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper talks about making sure that AI models are fair and don’t make mistakes based on small groups of people. They introduce a new way to do this called ExMap, which uses special diagrams from the model’s own thinking to figure out what it thinks each group is like. This allows them to train the model without needing labels for every single person in those groups. The results show that ExMap works really well and can be used with other methods to make AI models even better. |
Keywords
* Artificial intelligence * Clustering * Supervised * Unsupervised
