Article: Optimisation of academic gap compensation strategy based on transfer learning Journal: International Journal of Information and Communication Technology (IJICT) 2025 Vol.26 No.40 pp.55 - 69 Abstract: Identifying and addressing students' academic gaps are essential for delivering effective personalised learning experiences. In this study, we present a transfer learning model that combines transformer layers with convolutional modules to detect learning deficiencies and recommend targeted exercises. The model analyses student interaction data from an online homework platform, capturing patterns that indicate areas of misunderstanding. By integrating both global sequence modelling and local feature extraction, the system predicts performance outcomes with high accuracy. In experiments, the model achieved 87.5% accuracy and an AUC of 0.91, outperforming traditional approaches across multiple benchmarks. It also processes each student sequence in under 0.15 seconds, supporting its practical use in real-time learning environments. These results confirm the model's capability for accurate prediction, reliable gap detection, personalised intervention, and practical deployment in adaptive learning systems. Inderscience Publishers - linking academia, business and industry through research

Title: Optimisation of academic gap compensation strategy based on transfer learning

Authors: Yongmei Fu

Addresses: School of Marxism, Shandong Huayu University of Technology, DeZhou, 253034, China

Abstract: Identifying and addressing students' academic gaps are essential for delivering effective personalised learning experiences. In this study, we present a transfer learning model that combines transformer layers with convolutional modules to detect learning deficiencies and recommend targeted exercises. The model analyses student interaction data from an online homework platform, capturing patterns that indicate areas of misunderstanding. By integrating both global sequence modelling and local feature extraction, the system predicts performance outcomes with high accuracy. In experiments, the model achieved 87.5% accuracy and an AUC of 0.91, outperforming traditional approaches across multiple benchmarks. It also processes each student sequence in under 0.15 seconds, supporting its practical use in real-time learning environments. These results confirm the model's capability for accurate prediction, reliable gap detection, personalised intervention, and practical deployment in adaptive learning systems.

Keywords: academic gap detection; transfer learning; transformer; personalised learning.

DOI: 10.1504/IJICT.2025.149814

International Journal of Information and Communication Technology, 2025 Vol.26 No.40, pp.55 - 69

Received: 29 Jul 2025
Accepted: 17 Sep 2025
Published online: 13 Nov 2025 *

Title: Optimisation of academic gap compensation strategy based on transfer learning

Keep up-to-date