Abstract

The organisation of an Artificial Neural Network (e.g., the organisation in layers, the number of cells per layer, the degree of connectivity between the cells) has a big influence on its abilities (e.g., learning ability). In our work we present a novel method to organise the nodes and links of an Artificial Neural Network in a biologically motivated manner using virtual embryology. For this purpose we developed a virtual embryo-genesis, which mimics processes observable in biology. In our system a virtual embryo consists of individual cells, controlled by a genome. These cells can develop to nodes in the ANN during the embryogenetic process. The embryo is implemented as a spatially discrete and temporally discrete multi-agent model. The cells in our model interact with each other via virtual physics and via virtual chemistry. With the work at hand, we show that patterns developing in our virtual embryogenesis are comparable to patterns found during natural embryogenesis. We plan to combine the described virtual embryology with Evolutionary Algorithms to op-timise the genome of the embryo. We expect the described model of virtual embryology (in combination with Evolutionary Algorithms) to lead to novel, evolutionary shaped net structures of Artificial Neural Networks.