The aim of this project is to investigate and analyse different approaches for evolving weights of an artificial neural network (ANN). The report could be split in the following main sections:
- The first section looks at what Evolutionary Algorithms (EAs) are and investigates the different EA operators.
- The second part of the report performs statically rigorous analysis on the performed experiments and the gathered data. It uses mean, median, standard deviation and t-tests to justify why certain operators and parameters are better than others.
- The last section of the report, summarises the optimal operators and parameters that were found and their mean scores on both test and training datasets.
Please refer to the report for further details.
The following demo shows how a fleet of 8 spaceships are successfully landed. It is important to note, that they all have random initial positions and velocities:
