Summary of Learning From Noisy Labels Via Conditional Distributionally Robust Optimization, by Hui Guo et al.
Learning from Noisy Labels via Conditional Distributionally Robust Optimization
by Hui Guo, Grace Y. Yi, Boyu Wang
First submitted to arxiv on: 26 Nov 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 proposed conditional distributionally robust optimization (CDRO) framework learns from noisy annotations by estimating the true label posterior and minimizing the worst-case risk within a distance-based ambiguity set centered around a reference distribution. This approach addresses potential misspecification in the true label posterior, which can degrade model performances, especially in high-noise scenarios. The algorithm leverages the likelihood ratio test to construct a pseudo-empirical distribution, providing a robust reference probability distribution in CDRO. Experimental results on synthetic and real-world datasets demonstrate the superiority of this method. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper solves a problem with learning from noisy annotations, which is used to label large datasets. This noise can make it hard for models to learn accurately. The authors propose a new way to handle this noise using something called conditional distributionally robust optimization (CDRO). They show that by looking at the worst-case risk and finding a balance between being robust and fitting the data well, they can get better results than other methods. |
Keywords
* Artificial intelligence * Likelihood * Optimization * Probability
