Summary of Multi-group Proportional Representation in Retrieval, by Alex Oesterling et al.
Multi-Group Proportional Representation in Retrieval
by Alex Oesterling, Claudio Mayrink Verdun, Carol Xuan Long, Alexander Glynn, Lucas Monteiro Paes, Sajani Vithana, Martina Cardone, Flavio P. Calmon
First submitted to arxiv on: 11 Jul 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Information Retrieval (cs.IR); Information Theory (cs.IT); Machine Learning (cs.LG); 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 research paper proposes a novel approach to mitigate representational harms in image search and retrieval tasks. Current methods focus on balancing the number of retrieved items across binary attributes, neglecting intersectional groups defined by combinations of group attributes. The authors introduce Multi-Group Proportional Representation (MPR) as a metric to measure representation across these groups. They develop practical methods for estimating MPR, provide theoretical guarantees, and propose optimization algorithms to ensure MPR in retrieval. Experimental results show that optimizing MPR yields more proportional representation across multiple intersectional groups with minimal compromise in retrieval accuracy. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research paper helps make the internet a fairer place by fixing a problem in how images are found on search engines. Right now, some methods try to balance how many results they give for different groups of people based on just one or two things like gender and race. But this doesn’t work well for groups that have more than two characteristics, like people who are both women and Asian. The authors came up with a new way to measure how well search engines do at showing all these different groups fairly. They also showed that the current methods aren’t always good enough, and that their new approach can make things better. |
Keywords
» Artificial intelligence » Optimization
