|
| Author(s) | Gelenbe, E. (ed.) |
|---|---|
| Title | Neural networks II: advances and applications |
| Publisher | Elsevier |
| Year of publication | 1992 |
| Reviewed by | Anatoly Zhigljavsky |
Neural networks is an area of computer science dealing with construction and investigation of learning algorithms for recognition and other problems of artificial intelligence. Those interested in the subject may consult the journals "Neural Computation", "Cognitive Science", "Biological Cybernetics", as well as various books such as E. Kandel, J. Schwartz, Principles of neural science, Elsevier, 1985, or J. S. Judd, Neural network design and the complexity of learning, MIT Press, 1990.
The present volume contains 11 works devoted mainly to applications and various heuristics in construction of learning algorithms. Some papers seem to be competent, I could personally emphasis a paper of S. Shekhar, M. Amin, P. Khandelwal "Generalization perfomance of feed-forward neural networks".
The editor has divided the works into two groups of 6 and 5 according to whether they concern methodology or applications. I prefer another division which separates 4 articles authored by the editor. Compared with others, in average, these 4 articles are: (i) of higher theoretical level, (ii) typed in a word processor of a very low level, (iii) contain so many misprints that it is really hard to understand some of their formulas and propositions.
Certainly, specialists in learning theory will find interesting ideas in the volume. However, the production quality is rather low and the price seems to be unreasonably high. I would not recommend to buy this book.