Bayes-like Classifier with Fuzzy Likelihood

In this short paper we build a very simple classifier based on the concepts similar to Bayesian classifier using fuzzy theory. In our design we completely eliminate the concept of prior information about the class, and we just focus on the likelihood function (obtained from training data) and that is modeled as fuzzy sets. In the process of classification we have used the possibility-probability transformation. In this experimental study we show the efficacy of possibilitic description of knowledge for reasonable classification. The preliminary results obtained, in this study, by using fuzzy likelihood is as good as the Bayesian classifier.