Evaluating the Impact of Soft Computing Models on Student Performance in Intelligent Tutoring Systems

Abstract

This paper examines the influence of hybrid soft computing techniques—fuzzy logic, neural networks, and genetic algorithms—on student learning outcomes within intelligent tutoring systems (ITS). We compare three ITS configurations: fuzzy-only, neuro-fuzzy, and neuro-fuzzy with genetic optimization, across 200 high school mathematics learners. Results demonstrate that hybrid models significantly enhance learning gain (up to +25%), reduce error rates, and improve retention. The neuro-fuzzy-genetic system provides optimal ITS performance, suggesting a compelling avenue for adaptive educational technology.