Material Detail

This video was recorded at 27th AAAI Conference on Artificial Intelligence, Washington 2013. Incomplete-information games - such as most auctions, negotiations, and future (cyber)security settings - cannot be solved using minimax search even in principle. Completely different algorithms were needed. A dramatic scalability leap has occurred in our ability to solve such games over the last seven years, fueled largely by the Annual Computer Poker Competition. I will discuss the key domain-independent techniques that enabled this leap, including automated abstraction techniques and approaches for mitigating the issues that they raise, new equilibrium-finding algorithms, safe opponent exploitation methods, techniques that use qualitative knowledge as an extra input, and endgame solving techniques. I will finish by benchmarking poker programs against the best human poker professionals and by discussing what humans can learn from the programs.

Disciplines:

• Science and Technology  / Computer Science

More...