本文针对传统多层前向神经网络及其BP算法学习速度缓慢和泛化能力欠佳这两个缺点,分析了产生这些现象的原因,并且提出了一个崭新的改进模型—竞争式神经网络与相应的学习算法.这里所提出的竞争监督学习(有教师的学习),本质上是一个人机结合的学习过程.实验表明,新的算法使网络的学习和泛化能力有着明显的提高, In view of the disadvantages of traditional multi-layer forward neural networks and their BP learning algorithms, such as slow learning speed and poor generalization ability, this paper analyzes the causes of these phenomena and proposes a brand-new improved model - competitive neural network and corresponding The learning algorithm proposed here competition supervised learning (with teacher learning) is essentially a human-computer learning process.Experiments show that the new algorithm makes the network learning and generalization ability has been significantly improved,