Summary of The Role Of Learning Algorithms in Collective Action, by Omri Ben-dov et al.
The Role of Learning Algorithms in Collective Action
by Omri Ben-Dov, Jake Fawkes, Samira Samadi, Amartya Sanyal
First submitted to arxiv on: 10 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computers and Society (cs.CY); 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 This paper explores the concept of “collective action” in machine learning, where a group’s control over algorithms is studied. The authors argue that previous research has been limited by focusing solely on Bayes-optimal classifiers and neglecting the choice of learning algorithm. They propose an alternative approach using distributionally robust optimization (DRO) and stochastic gradient descent (SGD), two popular methods for improving worst-case performance. Empirical results demonstrate that the collective’s size and success depend heavily on properties of the chosen learning algorithm, underscoring the need to consider this factor in machine learning applications. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Collective action in machine learning means studying how a group can control algorithms. People used to think about how well a group does against Bayes-optimal classifiers, but that’s not enough. They didn’t consider what kind of algorithm is being used. In this paper, researchers look at two important algorithms: DRO and SGD. These methods are good for making sure the worst case doesn’t happen. The results show that whether a group is successful depends on the algorithm they’re using. This means we need to think about the algorithm when studying collective action in machine learning. |
Keywords
» Artificial intelligence » Machine learning » Optimization » Stochastic gradient descent
