Author(s): Mousumi Bala
Abstract: Autism spectrum disorder (ASD) is a complicated developmental disorder characterized by persistent difficulties in social interaction, speech and nonverbal communication, and repetitive activities. Because there is no medical test for ASD, diagnosing it might be difficult. ASD can be particularly difficult, with serious consequences for social interaction. In this study, we develop a machine learning model of human face recognition for autism children. We considered ORL database. For the face recognition, the data dimensionality reduction approaches such as principal components analysis (PCA), kernel principal component analysis (KPCA), independent component analysis (ICA) and factor analysis (FA) are investigated and compared. We have applied different classifiers and validated their performance through classification rate. We observed SVM showed the best performance with ICA methods for face recognition. This proposed system assists in the detection and recognition of human faces in autistic children.