基于前馈神经网络的模式识别效率分析

Research on the Network Training of the Feed-forward Neural Networks Algorithm

结合简易字符识别系统的设计过程,论述前馈神经网络算法的基本数学原理,并给出利用该算法求解模式识别问题的基本流程;重点研究利用BP算法进行网络训练的原理,并对影响网络训练效率的相关因素进行深入分析,结合MATLAB进行影响效应分析。通过实例仿真,结合仿真效果,验证前馈神经网络算法应用于求解模式识别问题的高效性。

Combining with the design procedure of the simple character recognition system, introduces the mathematics theory of the Feed-forward Neural Networks. Demonstrates the process that solves schema recognition and the theory of BP algorithm. At the same time, analyses the factors that can contribute to the networks training efficiency deeply by using MATLAB. Designs the character recognition system and confirms the algorithm is efficient in solving this type of problem.