摘要
为了全面优化BP神经网络,使之具有较好的泛化性能,改进并设计了一种遗传算法,并通过算法对比测试表明,改进后的遗传算法减少了内存占用量,保证了种群的多样性,提高了算法的运行速度和收敛效果。
In order to optimize completely BP neural network to make it have better generalization performance,this paper improves and designs a genetic algorithm,and through the comparison and test of algorithms,shows that the improved genetic algorithm reduces the memory footprint,guarantees the population diversity,and improves the algorithm running speed and convergence effect.
出处
《科技情报开发与经济》
2011年第1期119-121,共3页
Sci-Tech Information Development & Economy
关键词
BP神经网络
遗传算法
内存占用量
种群多样性
运行速度
BP neural network
genetic algorithm
memory footprint
population diversity
running speed
