机构地区: 广东工业大学计算机学院
出 处: 《东北电力学院学报》 2005年第4期51-55,共5页
摘 要: 首先对神经网络的数学模型和遗传算法的原理进行了系统阐述,然后在实验的基础上提出了对遗传算子的改进。重点讨论了采用十进制编码的遗传算法对神经网络进行函数按拟合的实现过程,总结了二进制编码的不足,并比较了改进前后的性能。 The paper presents a systematic description of both mathematical model of neural network and hereditary algorithm. Based on these theories, we put up some improvements of the existing hereditary operators. We specifically discuss the fitting process of hereditary algorithm with decimal coding and then analyze some considerate weaknesses of binary coding. We also evaluate the performatory differences between them.
领 域: [自动化与计算机技术] [自动化与计算机技术]