MurtadaMalik/Maze-Navigation-via-a-Differential-Evolution-and-Novelty-Search-Trained-Neural-Network

This code combines the differential evolution algorithm with novelty search to train a neural network controlled robot to navigate through a maze. Two mazes are implemented here. Novelty decay is implemented to aid the differential evolution algorithm's performance. Please use the following bib file to cite the paper describing this algorithm, presented at the ALA workshop at the 2018 Federated AI Meeting (ICML, AAMAS and IJCAI) in Stockholm, Sweden. @inproceedings{mason2018maze, title={Maze navigation using neural networks evolved with novelty search and differential evolution}, author={Mason, Karl and Duggan, Jim and Howley, Enda}, booktitle={Adaptive and Learning Agents Workshop (at ICML-AAMAS 2018)}, year={2018} }

JavaGPL-3.0