基于自适应动量因子的BP神经网络优化方法研究

Research on BP Neural Network Optimization Method Based on Adaptive Momentum Factor

人工神经网络是模仿动物神经网络行为并执行分布式并行信息处理的数学模型。网络依赖于系统的复杂性,调整大量节点之间的连接,达到处理信息的目的。因BP神经网络具有自适应性、自组织性和实时性等特点。目前,它广泛应用于模式识别、预测估计、信号处理等领域;因BP网络是基于梯度下降法实现算法学习的,所以不可避免地存在算法收敛效率较低的情况,非常容易停靠在局部最小点上导致在预测问题上效果一般。如何优化改进BP网络一直是一个备受关注的焦点。本文从两方面着手改进BP神经网络,并以在出版物中的图像识别为应用进行研究,以求提高网络收敛性和预测精度。

Artificial neural network(ANN) is a mathematical model that imitates the behavior of ANN and performs distributed parallel information processing. The network relies on the complexity of the system,adjusting the connection between a large number of nodes to achieve the purpose of processing information. Because BP neural network has the characteristics of self-adaptability,selforganization and real-time. At present,it is widely used in pattern recognition,prediction and estimation,signal processing and other fields. Because BP network is based on gradient descent method to realize algorithm learning,inevitably,the convergence efficiency of the algorithm is low,and it is very easy to stop at the local minimum point,which leads to the general effect on prediction problem. How to optimize and improve BP network has always been a focus of attention. In this paper,BP neural network is improved from two aspects,and the application of image recognition in publications is studied in order to improve the convergence and prediction accuracy of the network.