Summary of “normalized Stress” Is Not Normalized: How to Interpret Stress Correctly, by Kiran Smelser et al.
“Normalized Stress” is Not Normalized: How to Interpret Stress Correctly
by Kiran Smelser, Jacob Miller, Stephen Kobourov
First submitted to arxiv on: 14 Aug 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 paper explores the relationship between scaling and stress, a common quality metric used in dimension reduction projections of high-dimensional data. Stress is often employed to measure projection accuracy or faithfulness to the full data, but normalized stress can be sensitive to uniform scaling of the projection, despite this not meaningfully changing the projection’s properties. The authors investigate the effect of scaling on stress and other distance-based quality metrics both analytically and empirically. They introduce a simple technique to make normalized stress scale invariant and demonstrate its accuracy in capturing expected behavior using a small benchmark. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper looks at how making high-dimensional data smaller helps us understand it better. We often use special plots called scatter plots to show this data, but checking if the plot is accurate is tricky. That’s why we use quality metrics like stress to measure how good the reduced data is. One problem with using stress is that it can be affected by just scaling up or down all the points in the plot, even though that doesn’t change what the plot shows. The authors examine how this scaling affects stress and other similar metrics. They also come up with a new way to make stress less sensitive to scaling and show that it works well on some test data. |
