Summary of Data Valuation by Leveraging Global and Local Statistical Information, By Xiaoling Zhou and Ou Wu and Michael K. Ng and Hao Jiang
Data Valuation by Leveraging Global and Local Statistical Information
by Xiaoling Zhou, Ou Wu, Michael K. Ng, Hao Jiang
First submitted to arxiv on: 23 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
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 In this paper, researchers explore novel approaches to quantify the value of data within a corpus, crucial for machine learning tasks. They build upon Shapley value-based methods, which are widely used but often computationally expensive. The authors demonstrate that incorporating distribution information can significantly improve data valuation, proposing a new method that estimates Shapley values using both global and local statistical information. This approach is tested on various simulated and real datasets, showcasing its effectiveness in applications such as data removal, mislabeled data detection, and incremental/decremental data valuation. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Data valuation is important for machine learning tasks, but current methods can be slow or inaccurate. Researchers are working to improve these methods by using more information about the data’s value distribution. This paper shows how combining global and local statistical information can lead to better results in estimating Shapley values. The new method works well on different types of datasets and can even handle situations where data is added or removed over time. |
Keywords
» Artificial intelligence » Machine learning
