Simon Ellis, Graduate Student, Computer Science, RPI

 

Simon Ellis, Graduate Student, Computer Science, RPI

Sage 4101

November 11, 2015 12:00 PM - 1:30 PM

Abstract:

What makes a game a game? What makes a game easy or hard? Why do computers play some games well and most very badly? How could you make a AI more ‘interesting’?

The analysis of games and game-playing has long been a mainstay of research in the field of artificial intelligence (AI). Much of this work has been in a class of games readily tractable to game-theoretical approaches, such as chequers, othello and chess; many other games exist which by nature or design resist standard AI techniques. However, in 2011 a computer agent not only played but won at such a game: the game was Jeopardy!, and the computer was Watson.

In my talk I will discuss games and some of the issues which make them difficult for computers to play competently. After briefly describing the method of operation of IBM Watson, I will discuss my current research work on developing a cognitive game-playing system. I will conclude with a short demonstration of the current state of my prototype system Aleph, which isthe first board game AI to use the concepts and methods of cognitive computing in general and Watson in particular.

 

 

 

Add to calendar
Share|