Perceptual Associative Classifier based on Extended Fuzzy Logic

Rule base of a Fuzzy Associative Classifier (FAC) is a collection of fuzzy associative rules that are often based on the system's quantitative data. However, due to the real-world complexities and uncertainties, classification in many practical circumstances remains a matter of art of decision-making that is strongly influenced by the knowledge, experience, and personal perspective of individuals. In this paper, we introduce the f-associative fuzzy classifier (f-FAC) in the framework of Extended Fuzzy Logic (FLe), which is more closely related to the way of thinking and reasoning of human beings. In the proposed structure, human knowledge and experience are considered by fuzzy validity concept in both phases of construction and deduction of FACs. In this classifier, the validity of the items and rules is determined by integrating the opinion of experts on the basis of wisdom of crowds and using perceptual computing. To evaluate the proposed approach, a real dataset of burn patients in Ahwaz are considered. f-FARC-HD is then implemented as an extension of FARC-HD associative classifier and is compared with the other approaches (associative classifier and non-associative classifier). Also, f-FARC-HD and FARC-HD are compared in different levels of noise. Results indicate that considering the concept of validity in the proposed extended approach, f-FARC-HD, leads to comparable accuracy, but at a considerably less complexity. Also, f-FARC-HD is less sensitive against noise.