Statistics and Neural Networks: Advances at the Interface

Recent years have seen a growing awareness of the interface between statistical research and recent advances in neural computing and artifical neural networks. This book covers various aspects of current work in the area, drawing together contributions from authors who are leading researchers in the two fields. Their contributions show a strong awareness of the common ground and of the advantages to be gained by taking the wider perspective. Topics covered include: nonlinear approaches to discriminant analysis; information-theoretic neural networks for unsupervised learning; Radial Basis Function networks; techniques for optimizing predictions; approaches to the analysis of latent structure, including probabalistic principal component analysis, density networks and the use of multiple latent variables; and a substantial chapter outlining techniques and their application in industrial case-studies. This research interface is currently extremely active and this volume gives an authoritative overview of the area, its current status and directions for future research.