AUTHOR=Liu Wenbo , Li Ming , Zou Xiaobing , Raj Bhiksha 

TITLE=Discriminative Dictionary Learning for Autism Spectrum Disorder Identification

JOURNAL=Frontiers in Computational Neuroscience

VOLUME=Volume 15 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/computational-neuroscience/articles/10.3389/fncom.2021.662401

DOI=10.3389/fncom.2021.662401

ISSN=1662-5188

ABSTRACT=Autism  Spectrum  Disorder  (ASD)  is  a  group  of  lifelong  neurodevelopmental  disorders3with  complicated  causes.  A  key  symptom  of  ASD  patients  is  their  impaired  interpersonal4communication ability. Recent study shows that face scanning patterns of individuals with ASD5are often different from those of typical developing (TD) ones. Such abnormality motivates us6to study the feasibility of identifying ASD children based on their face scanning patterns with7machine learning methods. In this paper, we consider using the bag-of-words (BoW) model8to encode the face scanning patterns, and propose a novel dictionary learning method based9on dual mode seeking for better BoW representation. Unlike k-means which is broadly used in10conventional BoW models to learn dictionaries, the proposed method captures discriminative11information by finding atoms which maximizes both the purity and coverage of belonging samples12within one class. Compared to the rich literature of ASD studies from psychology and neural13science, our work marks one of the relatively few attempts to directly identify high-functioning ASD14children with machine learning methods. Experiments demonstrate the superior performance of15our method with considerable gain over several baselines. Although the proposed work is yet16too preliminary to directly replace existing autism diagnostic observation schedules in the clinical17practice, it shed light on future applications of machine learning methods in early screening of18ASD.