Neural networks: an engineering perspective

The development and application of neural networks are presented from an engineering perspective. It is stated that neural computing is a collection of mathematical techniques that have been gaining growing acceptance as plug-compatible replacements for statistical and other data-modeling techniques. Two of these techniques, function approximation and clustering, are discussed. The forces shaping the future of neural networking systems, including plug compatibility, hybrid systems-neural computing concepts integrated with expert systems, fuzzy logic, and genetic algorithms-and application specification systems, are reviewed.<<ETX>>