Hosagrahar Jagadish
University of Michigan, MI·Computer Science
Algorithms increasingly help decide who gets hired, admitted to college or approved for a loan. Many of these decisions rank people with a simple scoring formula, which is assumed to be easy to understand and justify.
- Fields
Computer Science
Social Science
- Topics
algorithmic fairness
algorithmic rankers
score-based ranking
- Grants
- 1 known grant, through Aug 2027
What they work on
- Develop methods to measure how item attributes and preprocessing choices affect the output of score-based rankers, to guide data scientists toward scoring functions that match their notion of quality.
- Quantify how data uncertainty and small changes to a scoring formula alter ranked outcomes, to guide data collection and the choice of stable ranking functions.
- Develop methods to measure unfairness in rankings toward disadvantaged groups under multiple fairness objectives and intersectional discrimination, and to help navigate trade-offs between fairness interventions.
More about the research
Yet even simple rankers can give counter-intuitive results. They can unfairly push down candidates from disadvantaged groups, and small changes in the data or the formula can reshuffle the outcome. Most fairness research has focused on classification, which labels one item at a time. Ranking compares items against each other, so it raises problems that classification methods cannot handle.
Why it matters
Many institutions rank people with simple scoring formulas, including employers, admissions offices and college-ranking publishers. Methods to check these rankers for instability and bias could help make such decisions fairer and more accountable, especially for candidates from disadvantaged groups. The work could also show when a ranker should be replaced by more data collection or another screening process.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Juba Ziani
Georgia Institute of Technology, GA·Computer Science·Early-career
Model how individual data providers shape the outcomes of machine learning systems through strategic choices about what data to share, combining tools from computer science and economics.
Vishnu Boddeti
Michigan State University, MI·Computer Science
Derive theoretical limits on the trade-offs among predictive accuracy, robustness across patient populations and privacy in healthcare AI models, and evaluate them in practice.
Sayan Bandyapadhyay
Portland State University, OR·Computer Science
Design polynomial-time approximation algorithms for balanced k-median and k-means clustering, where each cluster must fairly represent every protected group, aiming for constant-factor guarantees like those known for ordinary clustering.
The school
University of Michigan: 124 labs on PI Pages, $2.1B a year on research, 963 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.