Identificador persistente para citar o vincular este elemento:
https://accedacris.ulpgc.es/handle/10553/141826
Título: | Exploring Bengali handwriting as a tool for classifying children’s age groups | Autores/as: | Ismail, Shahid Diaz, Moises Aleman, Belen E. Ferrer, Miguel A. |
Clasificación UNESCO: | 33 Ciencias tecnológicas | Palabras clave: | Age Classification Bengali Handwriting Expert System Intervention School Going Children |
Fecha de publicación: | 2025 | Publicación seriada: | Multimedia Tools and Applications | Resumen: | Medical prescription, academic development, and content control are the most important age-related applications for school-going children. Among the aforementioned applications, the introduction of an intervention is critical for children who are lagging behind their peers in graphonomotor (handwriting and drawing) skills. Age classification using handwriting is an inverse problem which is related to the academic excellence of the children. It is inherently challenging in nature due to differences in learning abilities of growing children. To address the inherent problems in identifying the age of a child, we have put forth a machine learning-based system which is able to do multi-class classification. It separates the children of primary school into three age groups of 4, 6, and 9 years old, as this age group is most suitable for an intervention. We have introduced a decision system which can assist educators to assess the validity of a classification to introduce the intervention. This decision system utilizes a voting-based methodology for inference. Due to the detailed explanation that the decision system provides, it can act as a standalone or complementary part of an expert system for an intervention. We also utilized various deep learning-based methods and concluded that for unbalanced datasets, machine learning-based systems can outperform the deep learning-based systems. The computational load of machine learning is much less than deep learning-based systems due to the small number of features. Moreover, the best classification reported by our systems is 87% while using an artificial neural network, a classifier taken from machine learning domain. | URI: | https://accedacris.ulpgc.es/handle/10553/141826 | ISSN: | 1380-7501 | DOI: | 10.1007/s11042-025-20977-x | Fuente: | Multimedia Tools and Applications[ISSN 1380-7501], (Enero 2025) |
Colección: | Artículos |
Los elementos en ULPGC accedaCRIS están protegidos por derechos de autor con todos los derechos reservados, a menos que se indique lo contrario.