Applying a Neural Network Architecture with Spatio-Temporal Connections to the Maze Exploration
We present a model of Reinforcement Learning, which consists of modified neural-network architecture with spatio-temporal connections, known as Temporal Hebbian Self-Organizing Map (THSOM). A number of experiments were conducted to test the model on the maze solving problem. The algorithm demonstrates sustainable learning, building a near to optimal routes. This work describes an agents behavior in the mazes of different complexity and also influence of models parameters at the length of formed paths.
The reported study was supported by RFBR, research Projects No. 16-37-60055 and No. 15-07-06214.
- 1.Mnih, V., Kavukcuoglu, K., Silver, D., Graves, A., Antonoglou, I., Wierstra, D., Riedmiller, M.: Playing Atari with Deep Reinforcement Learning (2013)Google Scholar
- 2.Koutník, J., Šnorek, M.: Temporal Hebbian Self-Organizing Map for Sequences (2008)Google Scholar
- 3.Gupta, S., et al.: Cognitive Mapping and Planning for Visual Navigation. arXiv:1702.03920
- 4.Schrodt, F., et al.: Mario Becomes Cognitive. Top. Cogn. Sci. (2017). p. 131Google Scholar
- 5.Paxton, C., et al.: Combining Neural Networks and Tree Search for Task and Motion Planning in Challenging Environments (2017). arXiv:1703.07887
- 10.Chalita, M.A., Lis, D., Caverzasi, A.: Reinforcement learning in a bio-connectionist model based in the thalamo-cortical neural circuit. Biol. Inspired Cogn. Archit. 16, 45–63 (2016)Google Scholar