本文研究了手写体数字识别的新技术,采用模式识别传统技术与神经网络模型相结合的方法,即在抽取样本模式有效特征的基础上,训练神经网络分类器进行识别,所采用的神经网络分类器为带有一个隐层的多层网,它能在网络学习过程中自适应地调节隐元数。实验表明本系统的性能大大优于采用最近邻分类器的识别结果。本文研究的方法具有广义性,特别是自组织结构的神经网络分类器,可适用于其它模式识别任务。 In this paper, the new technique of handwritten numeral recognition is studied. The traditional method of pattern recognition is combined with the neural network model. On the basis of extracting the effective features of the sample pattern, the neural network classifier is trained to recognize. The neural network classifier As a hidden layer with a multi-layer network, it can adaptively adjust the number of hidden elements in the network learning process. Experiments show that the performance of this system is much better than that of the nearest neighbor classifier. The method studied in this paper has a generalized, especially self-organizing neural network classifier, which can be applied to other pattern recognition tasks.