Research Topics

Interested in algorithmic game theory (AGT) at the Technion and want to learn more? Here are - in a nutshell - several research topics I work on, with representative papers.

Algorithmic mechanism / contract / information design

Modern algorithms can’t be designed in a void; they constantly interact with self-interested, strategic agents, and are an integral part of economic markets. This raises new challenges and exciting opportunities. I’m interested in exploring algorithmic aspects of the following three branches, identifying differences and finding unifying similarities.

Mechanism/auction design:
The “input” comes from strategic agents.
[DRT’17] [EFFTW’21]

Contract design:
The “output” goes to strategic agents - I see this as a new frontier for AGT.
[DRT’19] [DRT’20] [ALST’21] [ADT’21] [Our tutorial]
Interestingly, a classifier is also a type of contract – it can incentivize self-improvement or gaming depending on its design.
[ADPTT’20] [GNETR’21]

Information design:
The powerful idea that persuasion can replace monetary incentives.
[FJMNTV’12] [BTZ’21] [BTXZ’21]

Robustness, simplicity and beyond worst-case analysis

Optimal algorithms or economic mechanisms tend to be complex and rarely used in practice; what are alternative, more useful notions of optimality and in what sense do they yield robust and simple mechanisms?

Interdependent values

Sometimes winning an auction can be a curse; this model sheds light on why and turns auction theory into a whole new ball game.
[RT’16][EFTZ’21][AT’21][Our tutorial]

Market equilibrium, pricing and fairness

There turn out to be deep connections between computational complexity on the one hand, and the existence of market equilibrium where prices equate supply and demand on the other.
Market equilibrium is also closely related to fair resource allocation, and thus also to fairness in algorithmic decision making.
[RT’15][BNT’19][T’17 - Whitepaper]

© 2021 Inbal Talgam-Cohen
Template design by Andreas Viklund