Neural Networks for State Evaluation in General Game Playing
Unlike traditional game playing, General Game Playing is concerned with agents capable of playing classes of games. Given the rules of an unknown game, the agent is supposed to play well without human intervention. For this purpose, agent systems that use deterministic game tree search need to automatically construct a state value function to guide search. Successful systems of this type use evaluation functions derived solely from the game rules, thus neglecting further improvements by experience. In addition, these functions are fixed in their form and do not necessarily capture the game’s real state value function. In this work we present an approach for obtaining evaluation functions on the basis of neural networks that overcomes the aforementioned problems. A network initialization extracted from the game rules ensures reasonable behavior without the need for prior training. Later training, however, can lead to significant improvements in evaluation quality, as our results indicate.
Unable to display preview. Download preview PDF.
- 1.Bader, S., Hitzler, P.: Dimensions of neural-symbolic integration — a structured survey. In: We Will Show Them: Essays in Honour of Dov Gabbay. Kings College Publications, pp. 167–194 (2005)Google Scholar
- 2.Clune, J.: Heuristic evaluation functions for general game playing. In: Proceedings of the AAAI National Conference on Artificial Intelligence, Vancouver, pp. 1134–1139. AAAI Press, Menlo Park (2007)Google Scholar
- 4.d’Avila Garcez, A., Zaverucha, G., de Carvalho, L.: Logical inference and inductive learning in artificial neural networks. In: Proceedings of the ECAI Workshop on Neural Networks and Structured Knowledge (1996)Google Scholar
- 5.Fawcett, T.: Feature Discovery for Problem Solving Systems. PhD thesis, University of Massachusetts, Amherst (1993)Google Scholar
- 6.Finnsson, H.: Cadia-player: A general game playing agent. Master’s thesis, School of Computer Science, Reykjavík University (2007)Google Scholar
- 7.Finnsson, H., Björnsson, Y.: Simulation-based approach to general game playing. In: Proceedings of the AAAI National Conference on Artificial Intelligence, Chicago, pp. 259–264. AAAI Press, Menlo Park (2008)Google Scholar
- 8.Genesereth, M., Love, N., Pell, B.: General game playing: Overview of the AAAI competition. AI Magazine 26, 62–72 (2005)Google Scholar
- 9.Hölldobler, S., Kalinke, Y.: Towards a massively parallel computational model for logic programming. In: Proceedings of the ECAI Workshop on Combining Symbolic and Connectionist Processing, pp. 68–77 (1994)Google Scholar
- 11.Kuhlmann, G., Dresner, K., Stone, P.: Automatic Heuristic Construction in a Complete General Game Player. In: Proceedings of the AAAI National Conference on Artificial Intelligence, Boston, pp. 1457–1462. AAAI Press, Menlo Park (2008)Google Scholar
- 12.Love, N., Hinrichs, T., Haley, D., Schkufza, E., Genesereth, M.: General Game Playing: Game Description Language Specification. Technical Report LG–2006–01, Stanford Logic Group, Computer Science Department, Stanford University, 353 Serra Mall, Stanford, CA 94305 (2006), http://games.stanford.edu
- 13.Schiffel, S., Thielscher, M.: Fluxplayer: A successful general game player. In: Proceedings of the AAAI National Conference on Artificial Intelligence, Vancouver, pp. 1191–1196. AAAI Press, Menlo Park (2007)Google Scholar
- 14.Hitzler, P., Bader, S., Witzel, A.: Towards a massively parallel computational model for logic programming. In: Proceedings of the IJCAI Workshop on Neural-Symbolic Learning and Reasoning (2005)Google Scholar
- 16.Towell, G., Shavlik, J., Noordenier, M.: Refinement of approximate domain theories by knowledge based neural network. In: Proceedings of the AAAI National Conference on Artificial Intelligence, pp. 861–866 (1990)Google Scholar