The maximum \(\Sigma\)-count as a decision operator (Q1109583)

The main approaches in pattern classification can be grouped into two main categories: the statistical and the syntactic pattern classification. This paper presents another method based on fuzzy set theory. At first a brief overview of the statistical method is given and especially the maximum likelihood decision rule. Then the maximum \(\Sigma\)-count operator is introduced which in contrast to the maximum likelihood decision rule is based not on `a priori' probabilities but on fuzzy labels. An algorithm that operates on a parallel fuzzy labelling file is presented along with a real implementation of the maximum \(\Sigma\)-count operator. The advantages and the versatility of this approach against the statistical one are emphasized and final conclusions are drawn.