Summary of Detecting Out-of-distribution Samples Via Conditional Distribution Entropy with Optimal Transport, by Chuanwen Feng et al.
Detecting Out-of-Distribution Samples via Conditional Distribution Entropy with Optimal Transport
by Chuanwen Feng, Wenlong Chen, Ao Ke, Yilong Ren, Xike Xie, S.Kevin Zhou
First submitted to arxiv on: 22 Jan 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 approach to out-of-distribution (OOD) detection leverages empirical probability distributions that incorporate geometric information from both training samples and test inputs. By modeling OOD detection as a discrete optimal transport problem, the conditional distribution entropy score function is introduced to quantify uncertainty in test inputs being OOD samples. This novel method inherits merits of distance-based methods while eliminating reliance on distribution assumptions and specific training mechanisms. Experiments on benchmark datasets demonstrate improved performance over competitors. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary When we train machine learning models, it’s common to get new data that’s different from what we trained with. This is called out-of-distribution (OOD) data. To detect OOD data, people usually use distance-based methods, but these can be tricky and not very good. In this paper, the authors suggest a new way to detect OOD data by using something called optimal transport. They create a special score called conditional distribution entropy that helps figure out if new data is OOD or not. This approach is better than previous ones because it doesn’t need specific training methods or assumptions about how the data looks. |
Keywords
* Artificial intelligence * Machine learning * Probability
