Biological neurons, or nerve cells, receive multiple input stimuli, combine and modify the inputs in some way, and then transmit the result to other neurons. Artificial neural networks are an attempt to emulate features of biological neural networks in order to address a range of difficult information processing, analysis and modelling problems. The principal class of ANNs are so-called feed-forward networks, but other types of ANN are for example recurrent neural networks. Among the feed-forward networks the most widely used approach is the multi-level perceptron (MLP) model. The application range is broad from non-linear regression to land cover change modelling. The aim of the topic is to introduce the principles of ANN and to understand and demonstrate its use in geospatial modelling.
In progress (GI-N2K)