Improving neural network performance on the classification of complex geographic datasets

Neural Networks are now established computational tools used for search minimisation and data classification. They offer some highly desirable features for landuse classification problems since they are able to take in a variety of data types, recorded on different statistical scales, and combine them. As such, neural networks should offer advantages of increased accuracy. However, a barrier to their general acceptance and use by all but `experts' is the difficulty of configuring the network initially.  This paper describes the architectural problems of applying neural networks to landcover classification exercises in geography and details some of the latest developments from an ongoing research project aimed at overcoming these problems. A comprehensive strategy for the configuration of neural networks is presented, whereby the network is automatically constructed by a process involving initial analysis of the training data. By careful study of the functioning of each part of the network it is possible to select the architecture and initial weights on the node connections so the constructed network is `right first time'. Further adaptations are described to control network behaviour, to optimise functioning from the perspective of landcover classification. The entire configuration process is encapsulated by a single application which may be treated by the user as a `black box', allowing the network to the applied in much the same way as a maximum likelihood classifier, with no further effort being required of the user.