论文部分内容阅读
前馈神经网络BP算法的改进方案中,对网络训练(学习)过程中学习率和惯性系数进行模糊自适应调节,以提高收敛速度,是一项很有效的措施。文中具体分析了如何根据设计者的先验知识确定模糊规则和隶属函数,并以三比特异或函数(或称奇偶分类)的实现为例,验证了这种算法的改进、加速了BP网络的学习过程。
In the improved BP algorithm of feedforward neural network, it is a very effective measure to adjust the learning rate and inertia coefficient fuzzy adaptively in the network training (learning) process to improve the convergence speed. This paper analyzes how to determine fuzzy rules and membership functions based on the priori knowledge of designers. Taking the implementation of three-bit exclusive OR function (or parity) as an example, the improvement of this algorithm is verified and the speedup of BP network learning process.