Deep knowledge tracing method improves interpretability and robustness
RegKT: Interpretable and Robust Deep Knowledge Tracing With IRT-Regularizer
Machine Learning
Summary
Tracking how well students learn over time is important for education, but many advanced computer models that do this are hard to understand and sometimes make mistakes on small sets of student data. The authors created a new technique that helps these models not only make fewer mistakes but also be easier to interpret by teachers and researchers. This makes it more practical to use deep learning in real classrooms. Their technique adds a regularizing rule inspired by a well-known educational model to improve both reliability and clarity.
What this means in practice
- •For educational technology developers: Build adaptive learning software that tracks student knowledge accurately while providing interpretable insights for instructors.
- •For edtech data analysts: Analyze student learning data with models less prone to overfitting on small datasets, improving reliability of progress reports.
Authors
Samuel Girard, Juan D. Pinto, Jill-Jênn Vie, Amel Bouzeghoub
Abstract
As deep learning models continue to advance, knowledge tracing models have achieved higher accuracy. However, these gains come at the cost of reduced interpretability, which is crucial for practitioners in educational settings to adopt new methodologies. Additionally, deep learning models are prone to overfitting, particularly when dealing with the small datasets that are common in educational applications. In this paper, we propose a novel regularization technique designed to enhance the robustness of deep-learning-based knowledge tracing models, while simultaneously improving their interpretability. Our method addresses both the interpretability and overfitting challenges, making it more feasible for real-world educational applications.