Summary of Thresholding Data Shapley For Data Cleansing Using Multi-armed Bandits, by Hiroyuki Namba et al.
Thresholding Data Shapley for Data Cleansing Using Multi-Armed Bandits
by Hiroyuki Namba, Shota Horiguchi, Masaki Hamamoto, Masashi Egi
First submitted to arxiv on: 13 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 paper presents an iterative method to efficiently identify and remove harmful instances from a training dataset, thereby improving model performance. Building upon Data Shapley, which evaluates instance contribution to model performance, the new approach utilizes the thresholding bandit algorithm to accelerate the process. Theoretical guarantees are provided for accurate selection of harmful instances with sufficient iterations. Empirical evaluations using various models and datasets demonstrate the method’s effectiveness in balancing computational speed and model performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper develops a way to quickly find and remove bad data points that can harm machine learning models. This is done by combining two existing ideas: Data Shapley, which says how much each piece of data affects the model, and a special algorithm called the thresholding bandit. The researchers show that this new method works well and is fast, even when used with different models and types of data. |
Keywords
* Artificial intelligence * Machine learning
