Reasoning in Adversarial Domains

Faculty

• Dana Nau
• V.S. Subrahmanian

• Tsz-Chiu Au
• Ugur Kuter
• Austin Parker
• Amy Sliva

• Stephen Smith

What We Are About                                              

Researchers in several different disciplines—AI, OR, Control Theory, and Economics—do research on adversarial decision-making. The approaches and techniques (game tree search, Markov decision processes, dynamic decision networks, and so forth) have substantial similarities, but the similarities and differences are ill-understood because the relevant research communities are nearly disjoint. Our goals are:

• to do groundbreaking theoretical work on adversarial reasoning;
• to implement our theories in compelling applications;
• to foster cross-disciplinary understanding.

Accomplishments                    

• Our DBS strategy was the best-performing non-master-slave strategy in the 2005 Iterated Prisoner's Dilemma with Noise competition
• Kriegbot is the world's first serious program to play kriegspiel chess
• Our Workshop on Decision Making in Adversarial Domains was held in May 2005
• Bridge Baron used our algorithms to win the 1997 world championship of computer bridge. It still is doing well: it was a finalist in the 2003 and 2004 world computer bridge championships
• Our work on game-tree pathology back in 1979 founded a subfield of AI research in which others are still actively doing research

Publications

• need to put a list here