Issues of Neurodevelopment in Biological and Artificial Neural Networks
Stephan K. Chalup · 2001
In this paper abstractions of neurodevelopmental mechanisms of biological nervous systems are interpreted as changes in the structure, learning parameters and input data of artificial neural networks. A review of neurodevelopmental mechanisms reveals that structural changes follow a typical timeline. We propose it as part of an algorithmic model of neurodevelopment and early learning. We further discuss the concept of incremental learning in artificial neural networks. We conclude that incremental learning can currently only model a small part of neurodevelopment in biology. Neurodevelopment should be modeled in the context of evolution.