摘要
目前广泛应用于神经网络优化的方法是反向传播(Back Propagation,BP),但是BP神经网络的全局搜索能力很有限。文中探讨了两种全局优化算法:遗传算法(Genetic Algorithm,GA)和模拟退火(Simulated Annealing,SA),以及它们和BP算法结合形成的优化算法,并且比较了它们在神经网络优化中的优缺点。
At present,Back Propagation(BP) is the most widely used optimization techniques for training neural networks, but it has limited ability to find global solutions.Two global optimizing algorithms ,genetic algorithms and simulated annealing algorithms, have been discussed in this paper, including the performance of the combination of optimizing algorithms and BP algorithms, and also compare their advantages and disadvantages for optimizing neural networks.
出处
《现代计算机》
2007年第2期30-32,共3页
Modern Computer
关键词
BP神经网络
遗传算法
模拟退火算法
全局搜索
BP Neural Network
Genetic Algorithms
Simulated Annealing Algorithms
Global Search