Biomedical Engineering Reference
In-Depth Information
7.3 Conclusion
The paradigm of associative-projective neural networks makes it possible to create
simple and highly productive hardware for the simulation of neural networks
(neurocomputers). Modern electronics make it possible to create neurocomputers
for the simulation of neural networks containing hundreds of thousands of neurons
and hundreds of millions of interneuronal connections. One way to increase neu-
rocomputer speed is to create hardware for the not fully connected structure of
associative-projective neural networks.
References
1. Kussul E. M., Lukovitch V. V., Lutsenko V. N., Multiprocessor Computers for Control of
Transport Robots in Natural Environment,
USiM
, No. 5, 1988, pp. 102-104.
2. Drumheller M., Poggio T., On Parallel Stereo, Proceedings of IEEE Conference Rob. and
Autom., San Francisco, Calif., Apr. 7-10, 1986, 3, pp. 1439-1448.
3. Hopfield J.J. Neural Networks and Physical Systems with Emergent Collective Computational
Abilities, Proc. Net. Acad. Sci. USA, 1982, 79, pp. 2554-2558.
4. Webster W.P. Artificial Neural Networks and their Application to Weapons,
Naval Engineers
Journal
, 1991, pp. 46-59.
5. Kussul E. M., Rachkovskij D. A., Baidyk T. N., 1991, Associative-Projective Neural Net-
works: Architecture, Implementation, Applications. In: Neural Networks & Their Applica-
tions, Nimes, France, EC2 Publishing, pp. 463-476
6. Baidyk T. N., Neural Network Subsets to Resolve Some Problems of Artificial Intelligence,
Thesis for a Doctor's degree (D. Sc. Eng), Institute of Cybernetics of Ukraine, 28.09.1994.
7. Baidyk T. N., Neural Networks and Artificial Intelligence Problems. Kiev, Naukova Dumka,
2001, p. 264 (in Russian).
