A Novel Self Organizing Feature Map for Uncertain Data

Abstract

In real-world applications, sometimes there are uncertainties in the data set whether from the collection process or from the natural language. There are not many algorithms that can deal with this kind of data set. Therefore, in this paper, we develop a linguistic self-organizing feature map (LSOFM) that works with vectors of fuzzy numbers. The algorithm is an extension of the regular self-organizing feature map (SOFM). We found that the results from the LSOFM are similar to that from the SOFM. The results from the LSOFM can provide information that contains all the uncertainties from the input while the SOFM cannot.