Summary of Assessing Model Generalization in Vicinity, by Yuchi Liu et al.
Assessing Model Generalization in Vicinity
by Yuchi Liu, Yifan Sun, Jingdong Wang, Liang Zheng
First submitted to arxiv on: 13 Jun 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 This paper investigates the ability of classification models to generalize beyond their training data without relying on labeled test samples. Current approaches often calculate unsupervised metrics that correlate with out-of-distribution accuracy, but these can be affected by spurious model responses. To overcome this challenge, the authors propose incorporating neighboring test sample responses into the correctness assessment of individual samples. This vicinal risk proxy (VRP) method computes accuracy without relying on labels and consistently improves upon existing generalization indicators both theoretically and experimentally. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at how well computer models can make predictions when given new, unseen data. Right now, people often use special numbers to measure how good these models are, but this can be tricky because some models might just pretend they’re very confident or not confident at all. To fix this, the authors came up with a new way to look at the model’s answers that takes into account what other nearby samples say too. This helps give a better idea of how well the model really is. |
Keywords
* Artificial intelligence * Classification * Generalization * Unsupervised
