Auto-Label Threshold Generation for Multiple Relational Classifications based on SOM Network

Abstract

Classification and Association rule mining are two basic tasks of Data Mining. Classification rules mining finds rules that partition the data into disjoint sets. This paper is based on MrCAR (Multi-relational Classification Algorithm) and Kohonen's Self-Organizing Maps (SOM) approach. SOM is a class of typical artificial neural networks (ANN) with supervised learning which has been widely used in classification tasks. For small disjunction mining, we collocate with a new auto level threshold generation method in our algorithm to solve the problem of unclassified data of MrCAR. So, we optimize the classification rate of MrCAR with SOM network and improve the efficiency of classification. This approach is highly effective for classification of various kinds of databases and has better average classification accuracy in comparison with MrCAR. Finally the results convincingly demonstrated that our proposed algorithm has high accuracy.