A Brief Introduction to Neural Networks

Livro digital

Título:
A Brief Introduction to Neural Networks

Autor:
David Kriesel

Categoria:
Tecnologia > IA

Doador:
Raffaello D. N.

Sinopse:
Neural networks become less mysterious when their abstractions are traced back to the systems that inspired them. David Kriesel begins with the vertebrate nervous system, neurons, and receptor cells, then shows how technical neurons deliberately simplify biology. From there, the text establishes network topologies, activation order, bias units, training samples, learning curves, and gradient optimization with both conversational explanations and formal definitions. The core journey moves through the single-layer and multilayer perceptron, linear separability, error backpropagation, resilient backpropagation, and encoding problems. It then broadens the reader’s toolkit with radial basis functions, recurrent Jordan and Elman networks, Hopfield associative memory, learning vector quantization, self-organizing feature maps, and adaptive resonance theory. Excursuses connect these models to cluster analysis, prediction, and reinforcement learning. This is a patient but mathematically serious introduction for readers who want more than a surface tour of machine learning. Carefully designed illustrations, marginal notes, independent sections, and implementation references make it useful for structured study or selective consultation. The payoff is a comparative understanding of classic neural architectures: what each one represents, how it learns, and which kinds of problems it is built to solve.