论文部分内容阅读
提出了一种基于最小分类错误 (MCE)训练的采用多层感知器 (MLP)结构的模式分类器设计方法 .这是一种以分类错误率最小化为目标的模式分类器设计方法 ,将它用于 MLP分类器设计能够进一步提高分类器的性能 .采用 MLP实现 MCE训练中的分类损失计算 ,从而将 MCE训练过程与 MLP分类器设计统一在一个神经网络结构中 ,通过 BP算法予以实现 .这不仅能达到提高MLP分类器性能的目的 ,而且简化了它的设计过程
A design method of pattern classifier based on MCE (Multilayer Perceptron) is proposed, which is a pattern classifier designed to minimize the classification error rate For the MLP classifier design, the performance of the classifier can be further improved.Using MLP to realize the classification loss calculation in MCE training, the MCE training process and MLP classifier design are unified in a neural network structure and implemented by BP algorithm. Not only can achieve the purpose of improving the performance of MLP classifier, but also simplifies its design process