Neuroevolution for General Video Game Playing
Authors:
- 1 August 2015
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 200-207
Abstract
General Video Game Playing (GVGP) allows for the fair evaluation of algorithms and agents as it minimizes the ability of an agent to exploit apriori knowledge in the form of game specific heuristics. In this paper we compare four possible combinations of evolutionary learning using Separable Natural Evolution Strategies as our evolutionary algorithm of choice; linear function approximation with Softmax search and e-greedy policies and neural networks with the same policies. The algorithms explored in this research play each of the games during a sequence of 1000 matches, where the score obtained is used as a measurement of performance. We show that learning is achieved in 8 out of the 10 games employed in this research, without introducing any domain specific knowledge, leading the algorithms to maximize the average score as the number of games played increases.Keywords
This publication has 10 references indexed in Scilit:
- Human-level control through deep reinforcement learningNature, 2015
- The 2014 General Video Game Playing CompetitionIEEE Transactions on Computational Intelligence and AI in Games, 2015
- A Neuroevolution Approach to General Atari Game PlayingIEEE Transactions on Computational Intelligence and AI in Games, 2014
- A video game description language for model-based or interactive learning2013
- HyperNEAT-GGP2012
- High dimensions and heavy tails for natural evolution strategies2011
- Autonomous Evolution of Topographic Regularities in Artificial Neural NetworksNeural Computation, 2010
- Coevolving Strategies for General Game Playing2007
- Real-Time Neuroevolution in the NERO Video GameIEEE Transactions on Evolutionary Computation, 2005
- Learning Without State-Estimation in Partially Observable Markovian Decision Processes1994