A neuro-fuzzy approach to diagnose and classify learning disability

Abstract

The aim of this study is to compare two supervised artificial neural network models for diagnosing a child with learning disability. Once diagnosed, then a fuzzy expert system is applied to correctly classify the type of learning disability in a child. The endeavor is to support the special education community in their quest to be with the mainstream. The initial part of the paper gives a comprehensive study of the different mechanisms of diagnosing learning disability. Models are designed by implementing two soft computing techniques called Single-Layer Perceptron and Learning Vector Quantization. These models classify a child as learning disabled or nonlearning disabled. Once diagnosed with learning disability, fuzzybased approach is used further to classify them into types of learning disability that is Dyslexia, Dysgraphia, and Dyscalculia. The models are trained using the parameters of curriculum-based test. The paper proposes a methodology of not only detecting learning disability but also the type of learning disability.