Neural Network Structure Optimization through On-line Hardware Evolution

Most neural network models base their ability to adapt to problems on changing their interconnec-tion strengths according to a learning algorithm. Evolutionary technics and a special class of learning algorithms enable a neural network to have a dynamic structure too. While in the rst case we obtain an optimized a-priori architecture the latter allows on-line adaptation. However, most of those algorithms are computationally intensive and diicult to implement in hardware. This paper describes a fully digital implementation of a neural network with on-line automatic structure optimization.