Harvard EconCS Group
School of Engineering and Applied Science
Harvard University
Publications :: About :: People :: Research Areas :: Conferences :: Related Groups :: Seminar Schedule
|
|
Harvard EconCS Group
School of Engineering and Applied Science Publications :: About :: People :: Research Areas :: Conferences :: Related Groups :: Seminar Schedule |
The EconCS group is pursuing research, both theoretical and experimental, at the interface between computer science and economics. We draw on methodologies from AI, multi-agent systems, microeconomic theory, optimization and distributed systems. In particular, we are interested both in the design of electronic auctions, mechanisms and markets and in the constructive use of economic methodologies and economic theory within computational systems. A central challenge is to resolve conflicts between game-theoretic and computational constraints. Current topics of interest include: incentive-based environment design; online mechanism design for systems with dynamic agent arrivals; the design of mechanism infrastructures and currencies for distributed and peer-to-peer systems; the preference elicitation properties and design of indirect mechanisms; applications to e-commerce, including combinatorial exchanges; cryptographically secure auctions; network formation games; incentive-compatible distributed optimization; and generative models of economic networks.
(top)
Faculty
Former Ph.D. Students
Professor David C. Parkes
Current Ph.D. Students
Ruggiero Cavallo
David Chen (STM)
Florin Constantin
Jacomo Corbo
Shaili Jain (Co-advised with Prof. Michael Mitzenmacher)
Laura Kang
Sébastien Lahaie
Benjamin Lubin
Jolie Martin (STM)
Katy Milkman (STM)
Chaki Ng (Co-advised with Prof. Margo Seltzer)
Malvika Rao
Sven Seuken
Jeffrey Shneidman
(Co-advised with Prof. Margo Seltzer)
Chris Thorpe (Co-advised with Prof. Michael Rabin)
Haoqi Zhang
Masters Students
John Lai (part-time), A.B. 2005
Current Undergraduate Students
Adam Juda (Ph.D. ITM, 2007)
Jason Woodard (Ph.D. ITM, 2006; Assistant Prof. at Singapore Management University)
Visitors
Johan Pouwelse, Prof. at Delft Inst. of Technology (Summer 2007)
Takayuki Ito, Prof. at Nagoya Inst. of Technology (2005)
Jonathan Bredin, Prof. at Colorado College (Fall 2004)
Rajdeep Dash, PostDoc at University of Southampton (Fall 2004)
Loan Le, Ph.D. student at GMU (Spring 2004)
Debasis Mishra, Prof. at ISI in New Dehli (Summer 2003)
Trading Agent Competition
2005 Team: Ariel Kleiner, Evan Sprecher
2004 Team: Hassan Sultan, Lukasz Strozek, Qicheng Ma, and David Hammer.
2003 Team: Rui Dong, Wilfred Yeung, and Terry Tai.
Former Undergraduate Students
Aaron Bernstein (A.B. candidate, MIT)
Andrew Bosworth (A.B. 2004, Facebook) [Master's Thesis]
Ryan Davies (A.B. 2005, McKinsey & Co.) [Master's Thesis]
Brad Diephuis (A.B., expected 2008)
Rui Dong (A.B. 2005, D.E. Shaw & Co.) [Master's Thesis]
Quang Duong (A.B. 2007, now Ph.D. student at U.Michigan)
Nick Elprin (A.B. 2005)
R. Kang-Xing Jin (A.B. 2006, Facebook) [Master's Thesis]
Ariel Kleiner (A.B. 2006, now Ph.D. student at UC Berkeley)
David Krych (A.B. 2003) [Master's Thesis]
DJ Lee (A.B. 2006, Google)
Qicheng Ma (A.B. 2006, Google) [Master's Thesis]
Sean MacLeod (A.B. 2003, Metacapital Management)
Ed Naim (A.B. 2004, Boston Consulting Group) [Master's Thesis]
Abe Othman (A.B. 2007, now Ph.D. student at CMU) [Master's Thesis]
Ivo Parashkekov (A.B. 2007, Boston Consulting Group) [Master's Thesis]
Aaron Roth (A.B. 2006 (Columbia), now Ph.D. student, CMU)
Aditya Sanghvi (A.B. 2006, McKinsey & Co.)
Saurabh Sanghvi (A.B. 2004, Goldman Sachs)
Grant Schoenebeck (A.B. 2004, now Ph.D. student at UC Berkeley)
Erik Schultink (A.B. 2007)
Jimmy Sun (A.B. 2007) [Master's Thesis]
Hassan Sultan (A.B. 2005, Bridgewater)
Aditya Sunderam (A.B. 2006, now Ph.D. student at Harvard)
Jie Tang (A.B., expected 2008)
Dimah Yankovsky (A.B. 2005)
Mark Yetter (A.B. 2007, Microsoft/Massive Inc)
Haoqi Zhang (A.B. 2007, now Ph.D. student at Harvard) [Master's Thesis]
(top)
| Algorithmic Mechanism Design | Work in algorithmic mechanism design classically focuses on the complexity of centralized implementations of game-theoretic mechanisms for distributed optimization problems. There are many interesting open problems, that at a high-level must resolve both computational and economic constraints. |
| Combinatorial Exchanges | Combinatorial exchanges (CEs) allow multiple buyers and sellers to submit bids, asks, and swaps on items. CEs hold promise for the reallocation of wireless spectrum, airport takeoff and landing slots, and grid computing resources. |
| Design with Bounded-Rational Agents | Mechanism design clasically assumes that agents are perfectly rational and able to play game-theoretic equilibrium strategies. In practice, agents can be resource-bounded (computationally and informationally) and may need to adjust towards an equilibrium or approximate an optimal strategy. This changes the framework for design. |
| Distributed Implementation | Mechanism design clasically assumes a trusted and powerful ``center", able to compute and enforce the outcome of social choice rules. Involving strategic agents in this computation is vital in distributed systems, but introduces additional strategic considerations. Distributed implementation seeks to bring the entire implementation--- information revelation and computation--- into an equilibrium. |
| Iterative (Combinatorial) Auctions | Iterative auctions are critical in settings with bidders that face hard valuation problems. An iterative auction can implement the efficient outcome without bidders revealing, or even computing, their exact values for all outcomes. Price-based auctions provide a particularly interesting and well-motivated subclass. |
| Online Mechanism Design | Many settings with self-interested agents are dynamic, with agents that arrive and depart across time and decisions to make on a rolling basis. This area of online MD seeks mechanisms that are incentive-compatible with respect to time and value, and implement sequential decision policies with good properties. |
| Preference Elicitation | Preference Elicitation considers indirect mechanisms in which agents provide incremental information about utility functions, with feedback guiding the elicitation process. There are many interesting open problems, including how to design second-best mechanisms in which incomplete information is unavoidable and how to characterize the minimal information requirements to implement particular social choice functions. |
| Systems and Infrastructure | Developments in the theory of Computational MD raise questions about how to deploy mechanisms in real distributed systems such as computational grids, peer-to-peer systems and sensor networks. Our systems and infrastructures work focuses on developing decentralized and scalable infrastructures for coordination and resource allocation. A plug-and-play market-based infrastructure with well-defined semantics is a core component in this vision. |
(top)
Seventh International Joint Conference on Autonomous
Agents and Multiagent Systems (AAMAS'08)
Ninth ACM Conference on Electronic Commerce (EC'08)
Twenty-third AAAI Conference on Artificial Intelligence (AAAI'08)
Third World Congress of the Game Theory Society (GAMES'08)
Third International Workshop on Internet and Network Economics (WINE'07)
INFORMS Annual Meeting (INFORMS'07)
Twenty-first International Joint Conference on Artificial Intelligence (IJCAI'09)
Past Workshops Organized by EconCS
Trading Agent Design and Analysis Workshop (July 2004)
Second Workshop on the Economics
of Peer-to-Peer Systems (June 2004)
Radcliffe Exploratory Seminar on Revealed and Latent Preferences: Economic and Computational Approaches (May, 2004)
Radcliffe Exploratory Seminar on Dynamic Networks: Behavior, Optimization and Design (October, 2006)
(top)
Artificial Intelligence Research Group at Harvard
Systems Research at Harvard
Harvard Economics Department
Science, Technology and Management Program
(top)