Autism spectrum self-stimulatory behaviours classification using explainable temporal coherency deep networks and SVM classifier / Liang Shuaibing

Autism spectrum disorder is a very common disorder. An early diagnosis of autism is essential for the prognosis of this disorder. The common method for diagnosis utilizes behavioural cues of autistic children. Doctors require years of clinical training to acquire the ability to capture these behavio...

Full description

Saved in:
Bibliographic Details
Main Author: Liang , Shuaibing
Format: Thesis
Published: 2022
Subjects:
Online Access:http://studentsrepo.um.edu.my/14424/1/Liang_Shuaibing.pdf
http://studentsrepo.um.edu.my/14424/2/Liang_Shuaibing.pdf
http://studentsrepo.um.edu.my/14424/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Autism spectrum disorder is a very common disorder. An early diagnosis of autism is essential for the prognosis of this disorder. The common method for diagnosis utilizes behavioural cues of autistic children. Doctors require years of clinical training to acquire the ability to capture these behavioural cues such as self–stimulatory behaviours. In recent years, the advancement of deep learning algorithms and hardware enabled the use of artificial intelligence technology to automatically capture self-stimulatory behaviours. Using this technique, the work efficacy of doctors can be improved. However, the field of self-stimulatory behaviours research still lacks large, annotated data to train the model. Therefore, the application of unsupervised machine learning methods is adopted. Meanwhile, it is often difficult to obtain good classification results using unlabelled data, further research to train a model that can obtain good classification results and at the same time being practical will be valuable. Nevertheless, in machine learning, the interpretability of the created model is also important. Hence, we have utilized the Layer-wise Relevance Propagation (LRP) method to explain the proposed model. In this work, the major innovation is utilizing the spatio-temporal continuity of close frames as a free form of supervision and setting a global discriminative margin to extract slow-changing discriminative self-stimulatory behaviours features. Extensive evaluation of the extracted features has proven the effectiveness of those features. Firstly, the extracted features are classified by the k-means method to demonstrate the classification of self-stimulatory behaviours in a completely unsupervised way. The conditional entropy method is used to evaluate the effectiveness of features. Secondly, we have obtained the state-of-the-art results by combining the unsupervised TCDN method with optimised supervised learning methods (such as SVM, k-NN, Linear Discriminant Analysis). These state-of-the-art results prove the effectiveness of the slow-changing discriminative self-stimulatory behaviours features.