神经网络在手写体字符识别中的应用研究

The Study of Neural Networks Applied to Handwritten Character Recognition

总结了神经网络被用于手写体字符识别分类中的应用情况,分析了所采用的反向传播神经网络、径向基函数神经网络、自组织神经网络等网络模型及其最新的应用情况,详细介绍了神经网络集成的一些方法.并根据目前的研究现状,指出用于分类器设计的神经网络集成方法和理论研究是未来的重要研究课题.

The Neural Network （NN） is extensively applied to the recognition of handwritten character, which has strong nonlinear mapping and parallel computing ability. In this paper, the BP NN, RBF NN, Kohonen NN, etc, are investigated their newest developments of application, which are used to design the classifiers of handwritten character. Some methods of combining multiple classifiers are also checked. Finally, it is presented that the implementation methods and theoretical analysis of NN ensembles are the issues valuable for future exploration in this area.