Predicting Developmental Degrees of Music Expression in Early Childhood by Machine Learning Classifiers with 3D Motion Captured Body Movement Data: A Recent Study
Researchers have continued to be intrigued by the interaction between a child’s developmental level of music and their musical output. Currently, one notable part will be to study such connection using a quantitative technique and to discover some predicted methodology to statistically repeat such interaction. The author of this work collected developmental features of musical expressions in early infancy from viewpoints of aspects of body movement and applied a machine learning-based classification algorithm to those feature quantities obtained from the participant children. In two studies, classification models were applied to the feature quantity for 3-, 4-, and 5-year-olds. capture. A three-way nonrepeated ANOVA was used to highlight developmental degree and extract feature quantity, and a statistically significant difference in the movement data analysed of the moving average of distance such as the pelvis and right hand, the moving average of acceleration such as the right hand, and the movement smoothness of the right foot was observed. After letting classifiers train with categorical variables of developmental degree evaluated by the author with simultaneously recorded video, the author classified the developmental degree of children’s musical expression using machine learning classifiers using feature quantities of motion capture data. The report of the author Multilayer Perceptron Neural Network is the best classifier, and Boosted Trees is the second best. The sensitivity result revealed that pelvic movement was closely associated to the degree of musical development. In addition to a thorough examination of kinetic feature amounts or an increase in training sample data, classifiers such as deep learning and others might be considered to improve classification accuracy.
Author (S) Details
Tokoha-University, Shizuoka, Japan.
View Book :- https://stm.bookpi.org/MPLLE-V7/article/view/2870