摘要
BP神经网络具有很好的拟合非线性函数的能力,但传统BP算法收敛慢,容易陷入局部极小值。不少研究人员从BP神经网络的初始权值和阈值着手,期望通过对初始权值和阈值的优化来提高BP神经网络的性能。对此提出一种改进BP神经网络的算法,使用狼群算法优化BP神经网络的初始权值和阈值,用数据训练BP神经网络后预测函数的输出。最后通过一个非线性函数来验证所提出模型的有效性。
Though BP neural network could fit nonlinear function well, the traditional BP algorithm has slow convergence speed, and it easily traps into the local minimum. Many researchers expect to improve the BP neural network's performance by optimizing the initial weights and thresholds. The article puts forward an improved BP algorithm. It optimizes the initial weights and thresholds of BP neural network with wolf colony algorithm, then predicts the output of functions after training the BP neural network with data. Finally, nonlinear functions are employed to illustrate the effectiveness of the proposed model.
出处
《科技创新与生产力》
2016年第1期56-58,共3页
Sci-tech Innovation and Productivity
关键词
BP神经网络
狼群算法
函数拟合
BP neural network
wolf colony algorithm
function fitting