Skip to main content Skip to navigation

Mastery Guided Non-parametric Clustering to Scale-up Strategy Prediction

Project Overview

The document explores the integration of generative AI in education, specifically focusing on its application in adaptive instructional systems (AIS) for middle-school math learning. It presents an innovative mastery-guided, non-parametric clustering method designed to enhance the prediction of student problem-solving strategies by leveraging extensive educational datasets. This approach aims to personalize learning experiences and improve their effectiveness by employing deep learning models, particularly Long Short-Term Memory (LSTM) networks. A significant emphasis is placed on understanding the symmetries in learners' strategies, which is crucial for increasing prediction accuracy and ensuring fairness across diverse skill levels. The findings indicate that this advanced method not only scales up strategy predictions but also contributes to fostering a more tailored educational environment, ultimately leading to better learning outcomes for students.

Key Applications

Mastery Guided Non-parametric Clustering for strategy prediction

Context: Middle-school math learning using Adaptive Instructional Systems (AIS)

Implementation: Developed a prediction model using LSTMs and non-parametric clustering techniques to identify and cluster student strategies based on mastery levels.

Outcomes: Achieved high accuracy in predicting strategies, improved fairness in predictions across different skill levels, and efficiently scaled predictions using a representative sample of the dataset.

Challenges: Computational expense and slow convergence of LSTMs with large datasets.

Implementation Barriers

Technical barrier

LSTMs are computationally expensive and slow to converge in large datasets.

Proposed Solutions: Utilizing non-parametric clustering to reduce the effective model complexity and improve the scalability of the prediction model.

Project Team

Anup Shakya

Researcher

Vasile Rus

Researcher

Deepak Venugopal

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Anup Shakya, Vasile Rus, Deepak Venugopal

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies