A Fast and High Efficient Neural Network Classifier
Yihong Dong · Jisuanji gongcheng · 2003
This paper presents a new neural classification algorithm based on the competitive neural network. Combining the competitive neural network and the hierarchical clustering, the new model classifies the objects first, then uses Hebb learning rule to connect the sub clusters which are activated, finally merges the same connected sub graph of the same output nerve cells after deleting the infirm links between the nerve cells in the recessive layer. Out of the noise influence, this neural network model can classify the clusters which have the random form or random size. With last learning and good result of the classification, the model is a good classifier of the multidimensional data.