论文部分内容阅读
提出了一种适用于模式识别的新型神经网络模型——局部有监督特征映射网络,描述了该网络的拓扑结构和学习算法,研究了网络的基本性能,最后将其应用到了质量控制图的模式识别中。理论研究和仿真实验表明,该网络结构简单、算法简洁,收敛速度快、识别精度高,适用于需要大样本训练、随机干扰严重的复杂模式的分类与识别。
A new neural network model suitable for pattern recognition is proposed, which is locally supervised feature mapping network. The network topological structure and learning algorithm are described. The basic performance of the network is studied. Finally, it is applied to the mode of quality control chart Recognized. Theoretical studies and simulation experiments show that the proposed network is simple in structure, simple in algorithm, fast in convergence speed and high in recognition accuracy. It is suitable for the classification and identification of complex patterns that require large sample training and severe random interference.