An approach to evolved neural systems is presented which, although preliminary in nature appears to be able to give a number of advantages over other approaches. The main feature of the approach is evolution under influence of environment. Additionally the more exact simulation of biological evolution is provided. Other features are the continual learning during functioning through self-refinement and self-adjustment and the rules for unsupervised termination of a system existence in the case of its incompetence. The proposition of classification is given for evolved neural systems developed with the presented approach. Two classes of evolved neural system are defined. The first class is the original method of use of genetic algorithm principles to evolution of neural system. The second one is the simulation of evolution of the natural nervous systems.
[1]
David E. Goldberg,et al.
Genetic Algorithms in Search Optimization and Machine Learning
,
1988
.
[2]
Christopher M. Bishop,et al.
Neural networks for pattern recognition
,
1995
.
[3]
John H. Holland,et al.
Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence
,
1992
.
[4]
Dana H. Ballard,et al.
Cortical connections and parallel processing: Structure and function
,
1986,
Behavioral and Brain Sciences.
[5]
W. Bechtel,et al.
Connectionism and the Mind
,
1991
.
[6]
James L. McClelland,et al.
Parallel distributed processing: explorations in the microstructure of cognition, vol. 1: foundations
,
1986
.