Playing Angry Birds with a Neural Network and Tree Search

In this paper, we introduce a method that combines a deep neural network and tree search for an Angry Birds AI agent. This neural network is trained first by supervised learning from another expert and then by reinforcement learning from self-play. Tree search enhanced by the neural network trained with supervised learning is used to strengthen the agent’s game play policy during reinforcement learning. To the authors’ knowledge, this is the first time that this approach is used to develop an Angry Birds AI agent. Our agent participates in the 2018 Angry Birds AI Competition and will be made available after the competition. The authors hope other researchers can gain some pieces of useful information from our findings and make deep learning more popular in Angry Birds AI Competition.

[1]  Eyke Hüllermeier,et al.  A CBR Approach to the Angry Birds Game , 2015, ICCBR.

[2]  Konstantinos Blekas,et al.  A Bayesian Ensemble Regression Framework on the Angry Birds Game , 2016, IEEE Transactions on Computational Intelligence and AI in Games.

[3]  Michael Fink,et al.  Angry-HEX: An Artificial Player for Angry Birds Based on Declarative Knowledge Bases , 2016, IEEE Transactions on Computational Intelligence and AI in Games.

[4]  David Silver,et al.  Deep Reinforcement Learning with Double Q-Learning , 2015, AAAI.

[5]  Demis Hassabis,et al.  Mastering the game of Go with deep neural networks and tree search , 2016, Nature.

[6]  Jude Shavlik,et al.  An Empirical Evaluation of Machine Learning Approaches for Angry Birds , .

[7]  Simon M. Lucas,et al.  A Survey of Monte Carlo Tree Search Methods , 2012, IEEE Transactions on Computational Intelligence and AI in Games.

[8]  Simon M. Lucas,et al.  Open Loop Search for General Video Game Playing , 2015, GECCO.

[9]  Demis Hassabis,et al.  Mastering the game of Go without human knowledge , 2017, Nature.

[10]  Simon M. Lucas,et al.  Fast Approximate Max-n Monte Carlo Tree Search for Ms Pac-Man , 2011, IEEE Transactions on Computational Intelligence and AI in Games.