Using neural networks modelling as motivation for alternative assessment practices in higher engineering education

Abstract

The human brain has about 100 billion neurons. These neural networks can be simulated in the science of artificial intelligence. Thus are these mathematical models in artificial intelligence based on their biological neural network counterpart. One can use these mathematical models to model learning. Neural networks are based on collections of nodes or neurons that are connected in a tree pattern to allow communication between them. A single node is a simple processor but a multilayered network with supervised training is capable of complex tasks. Learning can be divided into surface or deep learning. Surface learning is a low energy, low cognitive approach. Deep learning are recognized by, leaner's accepting personal responsibility, enjoying the experience of learning and the ability to identify where to apply learning in industry or future work. It is thus beneficial if the neural networks are stimulated to a deep, constructive learning approach. Assessment can be a good method to shape learning. This article argues that by shifting to an alternative assessment approach one can shift a learner's neural networks from surface learning to deep constructive learning.