Introduction To The Theory Of Neural Computation, Volume I

by John A. Hertz

A clear exposition of the theoretical aspects of neural computation. No specific background other than mathematics (multi-variate calculus, differential equations, and linear algebra) is assumed. Annotation copyright Book News, Inc. Portland, Or.

Genres: Neuroscience, Artificial Intelligence, Science, Computer Science, Textbooks, Nonfiction, Mathematics

327 pages, Hardcover

First published December 1, 1990

Reviews

[McKinley Paul · Rating 4 out of 5]

Great book. A very concise and clear introduction to physics based associative memory networks. My only gripe is that sometimes the notation and presentation can be a bit too compact. It is complete and not repetitive, so sometimes even though something correct is stated you have to do some work in the margins to believe it or understand it. Didn’t finish it but just got the basics down from it.

[Leonardo]

Carlos Selmo recomienda esto para SVM (lo cuál es un error, porque el libro no habla de eso).

[J · Rating 4 out of 5]

Deep and fast-moving. The early chapters are excellent, on the mechanics of associative memory systems like the Hopfield model. Later on, I found the exposition of the Boltzmann Machine, in particular, rather terse. But then I am not a mathematician. Still, by far the best theoretical introduction to neural networks I've come across.

[DJ]

old but good text on physics and NN recommended by Arbib