Interactive Visual Learning In Machine Learning: A Cognitive Learning Theories-Driven Approach

Abstract

This project developed a web-based interactive learning environment for machine learning concepts, integrating Multimedia Theory, Cognitive Load Theory, and Dual Coding Theory. Several design principles, aligned with these theories, guided the creation of four interactive visualisations—ROC curves (from data points and class distribution views) and gradient descent (for linear and polynomial regression). These visualisations were designed to enhance comprehension by reducing cognitive load and supporting dual coding. Future work will involve systematic usability testing and effectiveness evaluation, focusing on knowledge acquisition and learner satisfaction, to further refine these tools for enhanced educational outcomes.