Summary of Simultaneously Achieving Group Exposure Fairness and Within-group Meritocracy in Stochastic Bandits, by Subham Pokhriyal et al.
Simultaneously Achieving Group Exposure Fairness and Within-Group Meritocracy in Stochastic Bandits
by Subham Pokhriyal, Shweta Jain, Ganesh Ghalme, Swapnil Dhamal, Sujit Gujar
First submitted to arxiv on: 8 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Multiagent Systems (cs.MA)
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 Bi-Level Fairness approach for stochastic multi-armed bandits (MAB) addresses the limitations of existing methods by considering two levels of fairness. At the first level, it guarantees a minimum exposure to each group based on certain attributes. The second level ensures meritocratic fairness within each group by allocating pulls according to arm merit. By adapting a UCB-based algorithm, Bi-Level Fairness achieves anytime Group Exposure Fairness and individual-level Meritocratic Fairness within each group. This approach decomposes regret bounds into two components: regret due to anytime group exposure fairness and regret due to meritocratic fairness within each group. The proposed BF-UCB algorithm balances these regrets optimally, achieving an upper bound of O(√T) on regret, where T is the stopping time. Simulated experiments demonstrate that BF-UCB achieves sub-linear regret, provides better group and individual exposure guarantees compared to existing algorithms, and does not result in a significant drop in reward with respect to UCB algorithm. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In this paper, researchers propose a new way to make sure fair decisions are made when we don’t know which option is best. They call it Bi-Level Fairness and it works by making two types of guarantees: one for groups and another for individual items within those groups. This helps make sure that everyone gets a fair chance, even if some options are better than others. The team shows that their new approach can be used with an existing algorithm to get the best results while still being fair. |
