Customer Reviews

Neural and Adaptive Systems: Fundamentals through Simulations

5 star:		(4)
4 star:		(0)
3 star:		(2)
2 star:		(1)
1 star:		(0)

 

 

This has got to be the best textbook I have come across! The amount of thought and care in producing this book is phenomenal. It goes into details very thoroughly, and provides plenty of examples which would easily satisfy the very bright and the very slow. The math in back of all this is ACTUALLY EXPLAINED PROPERLY! Buy the book - it's worth every cent.

Topics explored in this book include signal processing, feature extraction, linear and non-linear modelling, temporal models and an array of connectionist learning paradigms. These topics are demanding, particularly mathematically, however I have taken a lot from this book where I have had limited success with others. The authors know a great deal about their subject, which has enabled them to take a holistic approach, which explains fundamentals before building them into powerful solutions. Equations, while frequently cited are explained descriptively and often boxed separately from the main text. Learning is greatly enhanced by the 200 interactive tutorials. The NeuroSolutions software (limited version free) is the best I've seen in terms of the number of architectures and data processing algorithms provided and comes with an excel add in! I searched Amazon long and hard for this book and was not disappointed - it sets a new standard for technical education.

I am quite conflicted in my thoughts on this book. The pluses are that it is comprehensive, thorough and comes with a useful pedagogical simulator. The minuses are that it is covers a vast range of subjects with a skimpy coat of mathematical and theoretical glue. When I found myself looking for more maths and theory, the text usually directed me towards the simulation. The simulations are good but the coding is invisible to the user due the range of NN widgets included (analogous to programming using the Lego Mindstorms visual programing tool). This is a downside for those hoping to learning coding techniques - there are not explored here - but will suit users that want a workbench to visualise concepts in the text. I confess that I became bored with the simulator about midway through the book.

The authors set themselves a formidable task in producing this book, and depending on the reader's needs thaey have either excelled themselves, or else have concocted a animaed dodo - interesting but near extinction.

This is the best book to learn adaptive signal processing. The authors teach you the spirit of adaptation thorugh simulation played by yourself. Moreover, readers can build their own system using the NeuralSolution. No matter you are a student or an engineer, this book is very useful.

The book covers the topics fairly well. Unfortunately, you won't ever use this book outside of a classroom. It seems like every page either has an illustration, or a box saying "run simulation 5.17 fom the CD, and here are 3 graphs from that simulation", or "Use the Excel plugin on the data from the CD". I suppose that is useful to those who can't follow the math. I believe it is also what causes the excessive cost of the book.

DON'T buy the book if you are looking for ways to include neural networks or machine learning into your existing applications. The book gives you formulas, and a workbench program that (once you figure it out) lets you customize it's own built-in formulas with its own data sets. It is not significantly useful when wanting to write your own adaptive system, nor does it have any significant code samples to learn from. You get the formulas and illustrations, figuring out how to apply them is up to you.

Generally I would agree with the reviewer who titled his review "Mindstorms meets NN; same strengths and weaknesses".

This is a decent book that provides some interesting insights into the subject matter. There are some very nice intuitive ideas that gives meaning to some of the math. Unfortunately, other parts of the text lack this intuition and seem like a presentation of network structure and algorithms more than a presentation of the ideas behind them.

I started my research in this area with a software engineering and management background with some intermediate signal processing experience on top. I also read Duda and Hart's Pattern Recognition book along with months of literature review on the subject before tackling this one. I must say that I am still confused on many points.

This is not an introductory text and will most certainly require the reader to seek out other references to fill in its gaps.

Muy buen libro y lo mejor es que trae buenos ejemplos los cuales ayudan bastante en el entendimiento de las redes neuronales. Por supuesto es una lectura complementaria a textos más avanzados si es que se quiere ahondar en el tema.