Understanding Neural Networks is a textbook and workbook that provides a unique interactive learning environment. With or without the aid of a classroom instructor, it allows students and other users to learn about neural networks while gaining practical, hands-on experience with all of the leading network models. Each model is presented as realistically as possible. Also included are chapter exercises and questions, many with illustrations.
The key feature of this workbook is the software. Available for PC-ATs and compatibles and the Macintosh, these disks contain a collection of full-featured commercial-quality simulators for the most important network paradigms. The user interface is graphic and easy to use, and the simulators are consistent across all networks. The simulators can also build and train significantly large networks, allowing users to construct networks on their own with data relevant to their problems.
Volume 1 covers learning, attractor networks, and hierarchical networks (including back-propagation networks). Volume 2 takes up temporal networks (including recurrent networks), self-organizing networks, higher-order networks, and such new directions in neural networks as fuzzy networks and evolutionary networks. Both volumes contain instructions on how to use the workbook, an introduction, appendixes, a table of random numbers, a glossary, a bibliography, and index.