Hybrid Adaptive Neuro-Fuzzy Inference System- Runge-Kutta Classification and Extracted the Eye Movement Features
Dyslexia disease is a most found learning disability problem over millions of children. It affects their reading ability which should be diagnosed accurately to treat them. Most research works confirms that, dyslexia can be diagnosed better with the help of eye movement of children. In our previous research method namely Fuzzy member function With Adaptive Neuro-Fuzzy Inference System -Support Vector Machine (FANFIS-SVM), dyslexia detection is performed accurately based on eye movement of children’s. However this works lacks from accurate decision-making in case of more similar feature values and also it leads to more complexity in case of presence of increased noise. These issues are solved in our proposed research method by introducing Hybrid Adaptive Neuro-Fuzzy Inference System- Runge-kutta (HANFIS-RK) classifier which can accurately predict the disease level based on eye movement. In this work, initially eye movement signals are captured from the diseased and also normal children. Then, these signals are pre-processed using Linear Discriminant Analysis (LDA) method. After pre-processing, Eye Movement Features (EMF) is extracted from the signals. From these signals more relevant features are selected to achieve increased classification accuracy rate by introducing Improved Artificial Bee Colony (IABC) Algorithm. Finally these features are learned and classified using Hybrid Adaptive Neuro-Fuzzy Inference System- Runge-kutta (HANFIS-RK) method. Experimentation is conducted in the Mat lab simulation environment and it is confirmed from the simulation outcome, proposed HANFIS-RK can accurately detect dyslexia disease than the existing research methods.