PERSONALIZED E-LEARNING SCHEME BASED ON LEARNER MODE PREDICTION USING DEEP BELIEF NETWORK WITH INTERACTIVE AUTODIDACTIC SCHOOL OPTIMIZATION

Abstract

In E-Learning Systems, the learning styles of scholars are detected automatically, that provides a solid method to educators to customize the education to be made accessible to scholars. Therefore, in this paper we have developed an effective personalized e-leaning approach by employing IAS (Interactive Autodidactic School) optimization based DBN (Deep Belief Network) classifier. The proposed personalized e-leaning approach generates the data from the log files of each learner entering online learning platforms like MOOC/MOODLE. Then certain essential features like residing time taken on each session, average reside time consumed on each learning objects (LO), individual learner credentials like Learner ID, session ID and course ID, and finally the Q-score generated with the help of feedback from, the learner. Using these features, the DBN is trained to predict the various modes of learning especially, active/reflective, sensing/intuitive, visual/verbal and sequential/global given by FSLSM (Felder-Silverman Learning Style Model). Further, the training procedure is enhanced here by employing IAS algorithm, where the weight matrix of DBN is chosen optimally to minimize the time required for basic procedure. Once the learning mode is found for each learner, the proposed model extracts the learning material for every learner based on their learning preference.