Spatial Interaction Modelling: Neural Network Methods and Global Optimization

The development of spatial interaction models is one of the most significant contributions of spatial analysis to social science literature. In more recent years, technological innovations in many areas have strongly influenced the research of spatial interaction modelling. The powerful and fast computing environment has brought many scholars to spatial interaction theory once again, by utilizing evolutionary computation to breed novel forms of spatial interaction models [see, for example Fischer and Gopal, (1994); Fischer, Hlavackova-Schindler and Reismann, (1999a and 1999b); Bergkvist, (2000); Reggiani and Tritapepe, (2000); Mozolin, Thill and Usery, (2000)] leading to neural spatial interaction models. Although the neural spatial interaction models have been inspired by neuroscience, they are more closely related to conventional spatial interaction of the gravity type than they are to neurobiological models.