Neural Network Design (2nd Edition)

Livro digital

Título:
Neural Network Design (2nd Edition)

Autor:
Martin T. Hagan, Howard B. Demuth, Mark H. Beale, Orlando De Jesús

Categoria:
Tecnologia > IA

Doador:
Raffaello D. N.

Sinopse:
If you are trying to understand how neural networks actually work, this book starts where confusion usually begins, with the basics of neuron models and network architectures, then moves straight into a chapter-by-chapter build-up that includes perceptrons, feedforward and recurrent networks, and an early illustrative example with Hamming and Hopfield networks. The table of contents shows a very deliberate progression: history and applications, then vector spaces and linear transformations, then learning rules such as Hebbian learning, Widrow-Hoff, and backpropagation, before moving into optimization, generalization, dynamic networks, competitive learning, radial basis networks, ART, and stability. That sequence makes the book feel both conceptual and practical, with theory tied to training behavior and design choices. What the reader gets is a rigorous foundation for designing, training, and evaluating neural networks, not just using them as a black box. The author team and the structured contents suggest a technical, multi-author academic text that rewards careful study, especially for readers who want the math, algorithms, and engineering tradeoffs behind classic neural network methods.