School / Research Seminar –
Game AI with Statistical Forward Planning
Given by
Prof Simon Lucas
School of Electronic Engineering and Computer Science
Queen Mary, University of London
16th March 2022, starting 4pm, Room A07
School of Computer Science
Abstract:
Statistical forward planning algorithms make use of the forward model of a game to search for action plans based on the reward profiles of the playouts. This provides powerful AI for many games (or problems in general that have fast and easily copied simulation models). Example algorithms include Monte Carlo Tree Search, and Rolling Horizon Evolution.
In this talk I will give a brief overview of the algorithms and demonstrate their ability to play a variety of games surprisingly well without the need for any prior training. I will also mention recent advances and future directions, including sample-efficient parameter tuning and Monte Carlo Graph Search.
Short Bio:
Simon Lucas is a professor of Artificial Intelligence at Queen Mary University of London where he leads the Game AI Research Group.
He is the founding Editor-in-Chief of the IEEE Transactions on Games and co-founded the IEEE Conference on Conference on Games.
His research involves simulation-based AI and evolutionary algorithms applied to Game AI, working towards Artificial General Intelligence.
He was head of school (EECS) at QMUL before taking up a visiting research scientist position at Meta.