模式神经元网络的聚类方法研究

The Clustering Study of Pattern Neuron Network

为了更好地模拟人脑对事物的学习、认知过程,笔者提出了模式神经元网络的聚类规则和方法,从而完善了这种新型的神经网络模型。与现有的人工神经网络不同,模式神经元网络不需要反复迭代就能达到学习、识别、分类的效果。实验结果表明:与自适应共振理论相比,模式神经元网络的学习效率快,识别精度高,分类效果也比较好。

In order to imitate the learning and cognitive process of human brain better, the clustering rule and method of the pattern neuron network was proposed, and this work has further improved the new artificial neural network model. Different from other current artificial neural network, the pattern neuron network can achieve good performance effect in learning, recognition and classification without the help of reiterating. The test result shows that compared with the adaptive resonance theory, the pattern neuron network can offer faster learning efficiency, higher recognition precision, and better categorized effect.