Summary of On the Inflation Of Knn-shapley Value, by Ziao Yang et al.
On the Inflation of KNN-Shapley Value
by Ziao Yang, Han Yue, Jian Chen, Hongfu Liu
First submitted to arxiv on: 25 May 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 A novel data valuation method is proposed, which utilizes Shapley values from cooperative game theory to quantify the usefulness of each individual sample. The approach aims to address challenges in current methods, such as value inflation, where some samples with positive values can be detrimental. To mitigate this issue, a calibrated version of KNN-Shapley (CKNN-Shapley) is introduced, which calibrates zero as the threshold for distinguishing beneficial from detrimental samples. Through extensive experiments, the effectiveness of CKNN-Shapley in alleviating data valuation inflation and detecting detrimental samples is demonstrated. The approach is extended to practical scenarios such as learning with mislabeled data, online learning with stream data, and active learning for label annotation. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper introduces a new way to evaluate the importance of individual data points using Shapley values from game theory. The method helps identify which data points are helpful or harmful to machine learning models. Right now, some methods can overestimate the value of certain samples, making them seem more useful than they actually are. To fix this, the researchers developed a new version of their approach that adjusts for these biases. They tested it on many different types of data and showed it works well in various scenarios. |
Keywords
» Artificial intelligence » Active learning » Machine learning » Online learning
