Kirthevasan Kandasamy
University of Wisconsin-Madison, WI·Computer Science·Early-career
Machine learning systems need large datasets, but the data are often scattered across organizations that each hold only a piece. Sharing data could unlock discoveries, yet organizations hesitate without clear incentives.
- Fields
Computer Science
Mathematics & Statistics
- Topics
machine learning
data sharing
game theory
- Grants
- 1 known grant, through Jun 2030
What they work on
- Design data-sharing protocols that assign data-collection duties in a balanced way and give out shared data in proportion to each participant's contribution, using game-theoretic analysis.
- Develop protocols that ensure data exchanges between competing organizations are mutually beneficial and resist strategic behavior such as skipping data collection, withholding data, or submitting false data.
- Analyze multi-round collaborations in which participants exchange data over time, adapt their strategies, and are uncertain about the value of others' data, and determine how to divide the benefits of a jointly trained model.
- Evaluate the proposed protocols through simulations and through partnerships with Alzheimer's disease research consortia.
More about the research
Several questions follow. Is sharing always worth it? Do participants get benefits that match what they contribute? How can a system stop some participants from benefiting without contributing? This project uses game theory, the mathematics of strategic decision-making, to build a theoretical foundation for fair and robust data sharing and collaborative machine learning.
Why it matters
Groups that hold separate datasets, such as research consortia in health care, public policy, and education, often fail to pool data because the incentives are unclear or unfair. Protocols that reward contribution and discourage free-riding or false data could make collaboration more attractive. Such protocols could help groups like Alzheimer's disease research consortia combine their data to train better shared machine learning models.
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.
Bhaskar Ray Chaudhury
University of Illinois Urbana-Champaign, IL·Computer Science·Early-career
Develop fair methods for allocating and exchanging chores, tasks that cost the people who receive them, as in online labor markets, using continuous optimization, equilibrium analysis and combinatorics.
Subhonmesh Bose
University of Illinois Urbana-Champaign, IL·Engineering·Early-career
Determine whether agents can learn each other's preferences through repeated interactions in incentive design problems, using tools from game theory, control theory and machine learning.
The school
University of Wisconsin-Madison: 91 labs on PI Pages, $1.9B a year on research, 854 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.