Neural networks in industry; a case study
Paul Friedel · Annales de Physique · 1992
Neural networks have recently known a major development after an extinction period of fifteen years. This is in part due to the conjunction of advances in mathematical techniques often linked to statistical physics, to a better knowledge in the field of neurobiology and to the dramatic increase of computing power in the recent years. This paper gives a pragmatic view of a consistent set of neural network techniques as well as some illustrative applications. It is by no means a review of neural network techniques or applications. Based on our own experience in Philips, we attempt to show the specific contribution of industry to the development of algorithms. The role of academic research is explored in this context.