C++ Neural Networks and Fuzzy Logic

This book is excellent introduction to theory and practical application of neural networks and fuzzy logic. I know that for sure because I was totaly illiterate in that topic before I red it. This book requires some prerequisite knowledge from reader. Reader must have at least average undergraduate knowledge of discrete matematics,probability theory and matrix algebra. I saw from previous reviews that some "good" C++ programmers expects to learn all that mathematics from one 500 pages book. I think that would be impossible even for a perfect neural network.

Despite being used for decades controlling safety/time-critical non-linear manufacturing processes (and other applications), the neural networks/AI label evokes images from science fiction by the general media. Beyond the initially intimidating title, this book is of real use to researchers addressing complex problems, and those trying to use neural networks. Without resorting to hardwired neural networks, or MATLAB with neural networks toolbox, one can readily understand and use the included (dated now) C++ neural network code (C compiler excluded- but use Borland/Microsoft). The basic neural network steps include: model process (e.g. manufacturing), gather data, pre-process data, compile, link & run the neural network code, let it iterate & learn, analyse output data sets, then use to hard-code a PID-controller (say). Even a 486-66/Win3.1/16MB RAM CPU can readily handle 25 variable/1000+ data point (pultrusion) manufacturing process (1994). As I often tend to use fuzzy-pre processing, and C/C++ for hacking demonstrators, the combined treatment of fuzzy logic, C++ and neural networks works very well.Topics include: fuzzy logic introduction, constructing a neural network, C++ and object orientation, models and Adaptive Resonance Theory (ART), learning, self organization & resonance, backpropogation and non-linear optimization, Bidirectional Associative Memory (BAM), Fuzzy Associative Memory (FAM), and applications for financial modelling.Weaknesses include: the now dated and programming-novice unfriendly software, and the ultimately limited software- far better when you are experienced to use MATLAB with neural networks toolbox for developing fast, usable networks.Overall, the style is approachable, and the content readily understandable and usable by the typical professional engineer/graduate engineering student audience. This book includes much helpful annotated pseudo-code, examples, references, defined terms, and mathematical explanations.

This book explains the basics of programming Neural Network(s) in C++ in NON-TECHNICAL terminology, which is very important for understanding basic concepts. It lays heavy emphasis on "learning by doing" approach, which also makes this book very attractive.

Your C++ neural algorithm has brought me to an Asia-Pacific paper competition on Petroleum and I WON IT! Excelent object Orientation!

You will find yourself rereading paragraphs to extract all the information from them. Written with concise and precision terminology that eliminates excess baggage on the book. Straighforward and to the point, yet very understandable, even to the most inexperienced in the field of neural nets. C++ demonstrations are usefull for programmers as well as for people interested in developing processing algorithms using neural network and fuzzy technology. Overall a great intro, or suplement to the field. A must in every library on neural processing