|
||||||
|
||||||
| Undergraduate Honours Projects | ||||||
|
Carleton University - School of Computer Science Undergraduate Honours Project Fall 2011 Implementation of Real-time Neuro-evolution with Augmented Topologies Stuart McIlroy 
ABSTRACT This paper describes a recent addition to the neural network library called Neuro-evolution with augmented topologies (NEAT) and an addition to it, Real-time NEAT (RTNEAT). It then goes into detail on the components of the algorithm. RTNEAT being designed explicitly for machine learning in video-games. This paper then describes three different experiments that were built, XOR, pole balancing and a predator prey game written in Microsoft's XNA platform that are designed to showcase the ability of NEAT and RTNEAT. It analyzes the experiments, the problems, results and performance. |
||||||
|
