Classification of Autism Spectrum Disorder: Deep Learning Approach

Year 2019, Volume: 2 Issue: 1, 18 - 20, 15.07.2019

Sevdanur Genç , Duygu Bağcı Daş

Abstract

Abstract : Autism is a complex developmental disorder that manifests itself as life-long neuropsychiatric disorder in the first years of life, manifested by significant delays and deviations in the area of interaction and communication and restrictive interests. Autistic individuals may have problems in social skills, language development and behavior. These problems are usually communicating to other people, making friends and difficulties in doing what is said. It is estimated that beside genetic causes, environmental reasons are also effective in development of autism. Today it is certain that there is not a single factor that causes autism. Autism is a complex disorder that occurs when multiple factors come together. Nowadays, many researchers have worked on more effective solutions to these complex disorders. For this purpose, classification estimations have been made using machine learning methods on various data sets that have been used in the literature. Deep learning is an another approach that has expanded machine learning and artificial intelligence scope. Deep Learning is a special kind of machine learning. It learns the examined world in the form of hierarchical concepts that are nested, defining each concept as an association with simpler concepts. At this point, classifications become very strong and flexible. In this study, it has been analyzed the data sets of Autism Spectrum Disorder using deep learning based classification approach which is a sub-branch of machine learning. As a result of the analyzes, it has been observed that the deep learning approach in test data gives better results than the other machine learning methods.

Keywords

classification, machine learning, artificial intelligence, deep learning, autism spectrum disorder.

References

• [1] LeCun, Yann, Yoshua Bengio, and Geoffrey Hinton. "Deep learning." Nature 521.7553 (2015): 436.
• [2] Derin Öğrenme Uygulamalarında En Sık Kullanılan Hiper-parametreler: https://medium.com/deep-learning-turkiye/derin-ogrenme-uygulamalarinda-en-sik-kullanilan-hiper-parametreler-ece8e9125c4 [Erişim tarihi: 08.12.2018]
• [3] Thabtah, F. (2017). Autism Spectrum Disorder Screening: Machine Learning Adaptation and DSM-5 Fulfillment. Proceedings of the 1st International Conference on Medical and Health Informatics 2017, pp.1-6. Taichung City, Taiwan, ACM.
• [4] Thabtah, F. (2017). ASDTests. A mobile app for ASD screening. www.asdtests.com [Erişim tarihi: 20 Aralık 2017].
• [5] Thabtah, F. (2017). Machine Learning in Autistic Spectrum Disorder Behavioral Research: A Review. To Appear in Informatics for Health and Social Care Journal. Aralık 2017 (in press).
• [6] Autism Screening Adult Data Set: https://archive.ics.uci.edu/ml/datasets/Autism+Screening+Adult [Erişim tarihi: 08.12.2018].
• [7] Autistic Spectrum Disorder Screening Data for Adolescent Data Set: https://archive.ics.uci.edu/ml/datasets/Autistic+Spectrum+Disorder+Screening+Data+for+Adolescent [Erişim tarihi: 08.12.2018].
• [8] Autistic Spectrum Disorder Screening Data for Children Data Set: https://archive.ics.uci.edu/ml/datasets/Autistic+Spectrum+Disorder+Screening+Data+for+Children [Erişim tarihi: 08.12.2018].
• [9] Knime. Software available at: http://www.knime.org/ [Erişim tarihi: 08.12.2018].
• [10] Deep Learning Tutorial MNIST: https://www.knime.com/nodeguide/analytics/deep-learning [Erişim tarihi: 08.12.2018].
• [11] Corani, Giorgio. "Air quality prediction in Milan: feed-forward neural networks, pruned neural networks and lazy learning." Ecological Modelling 185.2-4 (2005): 513-529.
• [12] Bottou, Léon. "Large-scale machine learning with stochastic gradient descent." Proceedings of COMPSTAT'2010. Physica-Verlag HD, 2010. 177-186.
• [13] Friedman, Jerome H. "Stochastic gradient boosting." Computational Statistics & Data Analysis 38.4 (2002): 367-378.