Personalized Hybrid Education Framework Based on Neuroevolution Methodologies

Abstract

The future pedagogical systems need anthropocentric inclusive educational programs in which the goal should be adjustable according to the knowledge requirements, intelligence, and learning objective of each student. Prioritizing these needs, innovative AI methods are required to assist and ensure the making of conscious educational decisions, in terms of clear identification and categorization with high accuracy of various forms of skills and knowledge of each student. This paper proposes a neuroevolution emerging technique that combines the searchability of evolutionary computation and the learning capability of a hybrid artificial neural networks method. Specifically, the proposed growing semiorganizing neural gas (GsONG) is a practical AI methodology utilizing advanced clustering techniques to enhance the learning experience by categorizing the true abilities, skills, and needs of learners, in an inclusive differentiated learning framework. It is a neural network architecture that includes competing and cooperating neurons with an unstructured mode whereby a cooperation-competition process delimits the topological neighborhood of neurons in a grid to identify patterns for which their classes are not known. To optimize the above process, a heuristic method was used that investigates the space of an objective function by regulating the optimal topologies of neurons that form pathway segments in a semi-contemplative manner. Based on the extensive experiments and results obtained from the GsONG clustering approach, the proposed algorithm can compensate with high accuracy for difficulties in multicriteria grouping and differentiation of uncertainty structures such as in small or tiny data sets.