心理学
发展心理学
幼儿
检查表
情绪识别
叙述的
认知心理学
情感(语言学)
情感工作
社会心理学
情绪性
情绪分类
应用心理学
感知
作者
Eshanie Rathnasinghe,Thiyara Gunawardena,Binuki Mihara,Chathuni Abeysinghe,Lakmini Abeywardhana,Lokesha Weerasinghe,Naren Selvaratnam
出处
期刊:
[Institute of Electrical and Electronics Engineers]
日期:2026-07-01
卷期号:: 1-7
标识
DOI:10.1109/ecai69016.2026.11613643
摘要
Early identification of emotional dysregulation in children aged 4–6 is critical to preventing long-term behavioral and cognitive impairments. However, traditional diagnostic methods often rely on invasive digital technologies, such as facial recognition, which raise significant privacy concerns and lack contextual depth. This paper proposes a privacy preserving, multimodal framework designed to identify emotional states and dysregulation levels in Sri Lankan preschool children through a combination of artistic expression and use of emotion regulation checklist. The methodology integrates two distinct data streams. The first utilizes a Deep Learning pipeline where a Vision Transformer (ViT-L/16) extracts high-dimensional features from a custom dataset of 600 child-generated artworks. These features are optimized via Principal Component Analysis (PCA) and classified using a Support Vector Machine (SVM) with a Radial Basis Function (RBF) kernel to identify emotions such as happiness, sadness, and anger. To address the "black-box" nature of AI, the system incorporates Explainable AI (XAI) features, including confidence rates and feature heatmaps, providing transparency for educators. The second stream processes the Emotion Regulation Checklist (ERC) via a Linear SVM to categorize dysregulation into low, medium, and high levels. Correlation analysis confirms that the ERC Lability/Negativity subscale serves as a robust ground-truth validator, showing strong correlations (r > 0.50) with total dysregulation scores. Implemented as a web-based diagnostic tool using React.js and Python, the system provides longitudinal time-series visualizations for parents and teachers. By bypassing invasive biometric monitoring in favor of culturally appropriate, non digital inputs, this research offers a scalable and ethically sound solution for early emotional intervention in urban educational settings.
科研通智能强力驱动
Strongly Powered by AbleSci AI