Neural Networks in C++: An Object-Oriented Framework for Building Connectionist Systems

Rating:
Summary: Good beginner book for NN's not for OOP
Review: In short, this is a good book for an experienced C++ programmer who has no experience in Neural Networks.

For somebody who has no experience whatsoever in Neural Nets, this book is a nice primer. It won't get you a Ph.D. (or an A in your undergrad AI class, for that matter), but if you've never been introduced before and want an overview that explores the concepts enough to get you started and whet your appetite for more, this book is a good place to start. It is short and easy to read, while still having enough substance to prepare you for more thorough books.

A major downfall of this book is, as others have mentioned, that the code provided is of poor quality. This book is definately not a good place to learn C++, the book contains some obvious mistakes like function definitions with no declaration, etc. Many more errors in the code are, I suspect, a result of the age of this book, it was written prior to the ANSI C++ standard: syntax and logical structures have changed significantly since 1992. It is, however, nothing a good coder can't handle.

Rating:
Summary: Look elsewhere.
Review: Perhaps he really does understand linear separability and how it applies to the exclusive or problem in neural networks. However, Mr. Blum's ludicrous excuse for an explanation of this classic problem fails to demonstrate anything, including that he knows what he's talking about. Unfortunately this is an exemplar for the entire book.

Wiley should be ashamed for continuing to peddle this

Rating:
Summary: Run away
Review: This book must be my worst investment. The code is full of mistakes. The theory side is even worse. I bought this book for the code, to see how one does implement neural networks in an object-oriented manner - after reading the book i knew how not to implement! BAD!!!