Type-2 Fuzzy Logic based Systems for Adaptive Learning and Teaching within Intelligent E-Learning Environments

The recent years have witnessed an increased interest in e-learning platforms that incorporate adaptive learning and teaching systems that enable the creation of adaptive learning environments to suit individual student needs. The efficiency of these adaptive educational systems relies on the methodology used to accurately gather and examine information pertaining to the characteristics and needs of students and relies on the way that information is processed to form an adaptive learning context. The vast majority of existing adaptive educational systems do not learn from the users’ behaviours to create white-box models to handle the high level of uncertainty and that could be easily read and analysed by the lay user. The data generated from interactions, such as teacher–learner or learner–system interactions within asynchronous environments, provide great opportunities to realise more adaptive and intelligent e-learning platforms rather than propose prescribed pedagogy that depends on the idea of a few designers and experts. Another limitation of current adaptive educational systems is that most of the existing systems ignore gauging the students' engagements levels and mapping them to suitable delivery needs which match the students' knowledge and preferred learning styles. It is necessary to estimate the degree of students’ engagement with the course contents. Such feedback is highly important and useful for assessing the teaching quality and adjusting the teaching delivery in small and large-scale online learning platforms. Furthermore, most of the current adaptive educational systems are used within asynchronous e-learning contexts as self-paced e-learning products in which learners can study in their own time and at their own speed, totally ignorant of synchronous e-learning settings of teacher-led delivery of the learning material over a communication tool in real time. This thesis presents novel theoretical and practical architectures based on computationally lightweight T2FLSs for lifelong learning and adaptation of learners’ and teachers’ behaviours in small- and large-scale asynchronous and synchronous e-learning platforms. In small-scale asynchronous and synchronous e-learning platforms, the presented architecture augments an engagement estimate system using a noncontact, low-cost, and multiuser support 3D sensor Kinect (v2). This is able to capture reliable features including head pose direction and hybrid features of facial expression to enable convenient and robust estimation of engagement in small-scale online and onsite learning in an unconstrained and natural environment in which users are allowed to act freely and move without restrictions. We will present unique real-world experiments in large and small-scale e-learning platforms carried out by 1,916 users from King Abdul-Aziz and Essex universities in Saudi Arabia and the UK over the course of teaching Excel and PowerPoint in which the type 2 system is learnt and adapted to student and teacher behaviour. The type-2 fuzzy system will be subjected to extended and varied knowledge, engagement, needs, and a high level of uncertainty variation in e-learning environments outperforming the type 1 fuzzy system and non-adaptive version of the system by producing better performance in terms of improved learning, completion rates, and better user engagements.