AI Techniques for Game Programming (Game Development) (Paperback)

The first third of the book is the usual primer on C++ - I skipped this chapter, and only gave the book 4 stars because of this. C++ does not interest me (I'm a java person) plus as a book focusing on AI I don't expect an intro to C++/windows programming - I wish authors of these books would make better assumptions about the likely reader.

The second third is 'genetic algorithms', very easy to follow and good examples starting with the Travelling salesman (example code is C++, but I don't mind that). It explains the basics of 'encoding' the 'genes' and how to evolve them, it also deals with different approaches to evolving a gene, the author makes it fairly clear that one approach may work for one problem, but not another.

The last third of the book is Neural nets. This was the bit that I was mostly interested in - I wanted the genetic section, but it was neural nets that I have got stuck on in the past when trying to read other books. Following a nice simple introduction of Neural Nets came the stuff on implementing them, what a breeze, I wish I had read this book a while ago. It also goes on to combine neural nets and genetic algorithms to demonstrate things like path finding. The author does bypass tricky parts, there are a couple of points he provides the maths for in a reasonably simple way and makes it quite clear that you should do further reading if you want to know the theory behind these parts, I felt this was a plus though others may find it off putting.

The book is clearly aimed at those with little knowledge of the 2 main topics covered, and does an excellent job of bringing that audience up to speed and ready to do more hard core reading if they want to become gurus. If you already feel comfortable with these 2 elements of AI then this book is probably not for you, If like me you don't know much, but want to then I recommend this strongly.

I have studied neural nets and GA, but this is the most interesting book I've read so far. It is written in such a way, that you can follow even if you have a limited (A-level) maths background. The author gives a fair programming and math explanations where needed.
If you do have a high level of programming/math background, it is easy to skip the relevant chapters.

It is the best book that gives you the 'feeling' of when and how to use these techniques, his examples are fun and easy to follow.

Unlike the rest of the LaMothe's books, this one has only a limited (just 30 pages) background on windows, and only the necessary explaining of the code.

I feel this book is very inspiring for a starter on GA and NN.
It has also managed to put together very nicely all the concepts on NN and GA that I've read during the past 8 years.

Highly recommended