Comparison on Performance of Classification of Autism Spectrum Disorder Using ML and Deeplearning
Autism spectrum disorder issue is a condition identified with mental health that impacts how an individual perceives and associates with others, causing problems in social connection and correspondence. The disorder likewise incorporates constrained and repetitive patterns of conduct. Artificial Intelligence techniques are being applied to autism child dataset 292 instances, adolescent dataset 104 instances and adult dataset 704 instances, to find valuable concealed hidden patterns and to build prescient models for detecting its hazard. Four machine learning and deep learning classifiers are applied to compare the performance of the classification of autism with respect to confusion matrix, Receiver operating characteristics area, true positive rate, false positive rate, precision, recall, f measure, Matthews correlation coefficient (MCC),Precision recall area (PRC).The results of CustumNet, multilayer perceptron deeplearning architecture, SVM , K Nearest Neighbour and Random forest algorithms are compared. The predominance of Deeplearning and SVM over the familiar classification algorithms in identification of the autism spectrum disorder is recognized for a framework of future BCI systems.