ADHD and ASD Classification Based on Emotion Recognition Data

In this work, we focused on classification of the participants with Attention Deficit Hyperactivity Disorder (ADHD), Autism Spectrum Disorder (ASD) and typically developing children, based on their performances during an emotion recognition experiment that we developed. We prepared an experiment environment where participants were shown images of faces of people exhibiting certain emotions up to a certain strength and then they answered the question "What is the emotion of this person?". The response and response latency of the participants were recorded and used for the classification process. Before the classification step, in order to select the relevant images which are used as features in this work, ReliefF feature selection algorithm was used. Machine learning feature selection and classification algorithms were used on different definitions of the classification problem where the differentiation between two classes against each other or one class against the other two classes were aimed. The selected features (images shown) and the classification performance changed based on the classification problem definition.