Analysis and Improvement of Feed-forward network training: With Complete Implementation - Softcover

Inhaltsangabe

This Text starts with a detailed overview of the Applications of neural Networks, illustrating its importance. The current problems present in the existing training algorithms like Back-Propagation, Newtons algorithm and the popular Levenberg-Marquardt algorithm are reviewed. The use of Multiple Optimal learning factors are explored in the text, followed by its complete analysis and possible methods of improvisation. All the training algorithms are implemented in Visual Studio 2005, and tested with universally accepted data files. It is observed that the Improvement of Multiple Optimal Learning Factors suggested in this text, provides almost the same effectiveness of the Levenberg-Marquardt algorithm with much lesser computational cost.

Die Inhaltsangabe kann sich auf eine andere Ausgabe dieses Titels beziehen.