Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
411171 | Neurocomputing | 2007 | 7 Pages |
Abstract
This paper proposes a distributive genetic algorithm for the learning of neural networks (DGANN). To tackle several well-known problems for conventional genetic algorithms (GAs), a synergetic multi-operator multi-population mechanism is developed, incorporating an αα transformation crossover operator and mixed-crossover operators. The proposed algorithm is applied to both benchmark numerical examples and pattern recognition of blue-green algae in lakes. Experimental results confirm that the proposed algorithm is superior to conventional GAs in terms of the convergence speed and solution precision, and is also capable of generating neural networks with significantly improved generalization performance.
Keywords
Related Topics
Physical Sciences and Engineering
Computer Science
Artificial Intelligence
Authors
Zhihong Yao, Minrui Fei, Kang Li, Hainan Kong, Bo Zhao,