The basic neural network topologies and learning principles are covered in clear, extensive detail in this book by the creators of the Neural Network Toolbox for MATLAB.
In it, the authors place an emphasis on a cogent description of the main neural networks, training techniques, and applications to real-world issues. Features extensive discussion of feedforward networks (including multilayer and radial basis networks) and recurrent network training techniques.
The backpropagation algorithm's conjugate gradient and Levenberg-Marquardt variants are covered in the text, along with Bayesian regularization and early halting, which guarantee trained networks' generalizability. Simple building blocks are used to describe feature maps, learning vector quantization, and associative and competitive networks.
There is a chapter with helpful training advice for function approximation, pattern recognition, clustering, and prediction, and there are five chapters with in-depth case studies from real-world scenarios. several solved issues and thorough illustrations.
