Fuzzy logic and deep learning integration in likert type data

In recent years, the deep learning approaches, which have been developed from the Artificial Intelligence techniques based on the way that the human brain works and which are built on the principles of artificial neural networks, have gained great importance because it is an effective method in the recognition and classification of data. The aim of this thesis is to analyze the performance of deep learning techniques in the form of a 5-point Likert-type scale by converting the artificial data sets into a fuzzy form using triangular or trapezium fuzzy numbers. In order to test the performance of the proposed model which is the integration of deep learning and fuzzy logic techniques, the satisfaction estimation problem involving uncertainty and complex relations was chosen. The data sets generated by fuzzy numbers are at least 3 or 4 times variables more than the normal data set. In the optimization studies with this data, the problem of falling into the local optimum is eliminated and it reaches to the global optimization point. In this study, first of all pilot study was done by using a classical method such as logistic regression which forms the basis of many classification techniques. Then, the performance of deep learning technique which is in the center of attention due to the advantages, has been evaluated in the framework of fuzzy logic. In the analysis conducted with deep learning, it has been clarified with separate results for peak, maximum and minimum values which constitute an example of blurring in literature. In contrast to the literature, it was suggested that fuzzy numbers produce a single result sequence and the performances of the deep learning model were investigated. Also, it is thought that this study will contribute to Turkish literature because of the limited literature in Turkish about deep learning. Keywords: Artificial Intelligence, Deep Learning, Fuzzy Logic, Likert Scale