Advanced SOM Algorithm Based on Extension Distance and Its Application

摘要

In order to solve the low efficiency problem of traditional SOM, a novel model is proposed based on the self-organized map neural network by using the extension theory. A novel extension distance is introduced and aimed to calculate the similarity of data points from the class domain. A proposed extension distance with a distance parameter is used to make the procedure of clustering controlled. It is shown that the proposed advanced SOM based on extension distance has a faster learning speed when compared with SOM neural networks; moreover, the new model is proved to have higher accuracy and lower cost of memory. It is an improvement of the traditional SOM. The new model is testified in respect of its effectiveness and feasibility in experiment on two different datasets.