Tuesday, November 12, 2024, 12pm

Machine learning (ML) models increasingly make or inform decisions in domains such as financial markets, e-commerce, and fintech lending. Due to the high-stakes nature of the decisions being made, ML-based systems operating in these domains need to be able to reliably reason about different outcomes in the presence of various forms of strategic behavior. In particular, the principal (i.e. the person/institution deploying the ML model) often needs to reason about the incentives of other individuals in the market, leverage the information discrepancy between themselves and others, and learn causal relationships between observable features and outcomes of interest, often under partial feedback about the underlying learning/decision-making problem. Moreover, these issues often compound in realistic decision-making scenarios, requiring the principal to handle two or more of them simultaneously. In this thesis proposal, I will highlight the research I have done so far to address the additional challenges which come with learning and decision-making in online markets, and I will overview exciting directions for future research.

Thesis Committee
Nina Balcan (Co-chair)
Steven Wu (Co-chair)
Tuomas Sandholm
Michael I. Jordan (University of California, Berkeley / INRIA Paris)

Additional Information

In Person and Zoom Participation.  See announcement.

Event Type: Thesis Proposals
Room Number: In Person and Virtual - ET
Building: Reddy Conference Room, Gates Hillman 4405
Speaker's Name: KEEGAN HARRIS
Speaker Website: keeganharris.github.io
Speaker's Professional Title: Ph.D. Student, Machine Learning Department, Carnegie Mellon University
Talk Title: Foundations of Data-Driven Decision-Making in Online Markets
For More Information: stidle@andrew.cmu.edu
Affiliations: Machine Learning Department (MLD)
Organization(s): School of Computer Science