NEATfields: Evolution of large neural networks

The basic structure of a NEATfields network.

In the last decades, many researchers have used evolutionary algorithms to adapt the topology and connection weights of recurrent neural networks for various control tasks. This has become a useful machine learning technique. Because handling large genomes is difficult, however, these neural networks typically contain only a few neurons. If the genome contains a recipe for construction of the network instead of the network itself, it can be much smaller. We have developed a method than can exactly do this, and performs very well on a number of different problems.

We take the well known NEAT method as starting point and extended it with a higher level structure: the neural field. Elements of a neural field are small NEAT networks. The fields themselves form a higher-level NEAT-like network. Our method, called NEATfields also contains further operators to connect and modify the building blocks of such a network.