Personalized prediction of student progress in moral education using multi modal learning

Area of Science:

• Education
• Computer Science
• Ethics

Background:

• Accurate, individualized prediction of student development in ethics education is crucial for social well-being and civic education.
• Current methods lack unbiased insight into individual moral development paths, hindering timely interventions.

Purpose of the Study:

• To propose a novel hybrid method for the individualized prediction of student development in ethics education.
• To enhance the accuracy and reliability of forecasting students' moral growth trajectories.

Main Methods:

• A four-step hybrid approach combining data preprocessing, Auto-Encoder neural networks for data encoding, a self-attention mechanism for modality fusion, and Random Forest for target variable forecasting.
• Utilized machine learning techniques including Auto-Encoders and self-attention for feature extraction and fusion.
• Employed Random Forest for final prediction based on aggregated, fused features.

Main Results:

• The proposed hybrid method demonstrated significant improvements in prediction accuracy.
• Achieved a 7.1% increase in accuracy and a 7% increase in F-measure compared to existing methods.
• Validated the effectiveness of the innovative machine learning and data fusion approach.

Conclusions:

• The hybrid method provides more accurate and reliable predictions of students' moral development.
• This approach supports the provision of appropriate and timely interventions in ethics education.
• The study highlights the potential of advanced machine learning for advancing educational psychology and practice.