An experimental study: using categorical or fuzzy inputs for classification problems with dimensionality reduction

A fuzzy control system is a mathematical framework that evaluates analog input data in terms of logical variables with continuous values ranging from 0 to 1. From the 1970s on, fuzzy notions have exploded in popularity across all fields. Fuzzy logic that contains fuzzy values, fuzzy variables, and fuzzy sets is frequently used by engineers, statisticians, and programmers to describe vague concepts mathematically. In recent years, researchers have begun to investigate fuzzy systems in deep neural networks. In our study, the use of fuzzy inputs and the use of categorical inputs in classification problems were compared with and without dimension reduction. For the combination of four different datasets and three different encoder pairs, the accuracy using fuzzy values was higher than the accuracy using categorical values, and a 4.54656% average increase in accuracy value is maintained. For big-data analysis, in critical fields like the medical field, even a tiny gain in accuracy can make a big difference in people's lives. I hope that the findings will guide researchers to consider the fuzzy representation of the data in the data pre-processing part.