Sign in Subscribe
Students

Leveraging Hybrid Deep Q-Learning for Early Identification of At-Risk Students

Strategic Insights

— 1 min read

Researchers at the Modern College of Business and Science have developed a new approach to predict student performance and identify at-risk students using a hybrid deep learning model. The proposed model, called Sea Lion Search Optimization (SLnSO), combines the benefits of Deep Q-Network (DQN) and other optimization techniques to improve the accuracy of predictions. The study suggests that the SLnSO-DQN model achieves significant performance based on various parameters, including Mean Absolute Error (MAE), Mean Square Error (MSE), and Root MSE (RMSE).

Key Takeaways:

  • The researchers employed a novel approach to predict student performance, incorporating external factors beyond learning activities, such as course duration.
  • The proposed SLnSO-DQN model leverages the strengths of Deep Q-Network (DQN) and other optimization techniques to enhance prediction accuracy.
  • The study achieved significant performance using various parameters, including MAE (0.327), MSE (0.265), and RMSE (0.514).
  • The researchers demonstrated the effectiveness of the SLnSO-DQN model in identifying at-risk students, aiming to help educators and institutions make timely interventions.
  • The study highlights the importance of considering external factors like course duration in predicting student performance.
  • The researchers employed the Damerau-Levenshtein technique for feature selection and Yeo-Johnson transformation for data preprocessing.
  • The study contributes to the ongoing development of predictive models for early identification of at-risk students.

Statistics:

  • Mean Absolute Error (MAE): 0.327
  • Mean Square Error (MSE): 0.265
  • Root MSE (RMSE): 0.514
  • The study involved a dataset with input data taken from the dataset and transformed using Yeo-Johnson transformation.
  • The researchers employed the Deep Q-Network (DQN) training algorithm for the SLnSO-DQN model.
  • The study also used Data Augmentation (DA) to increase the dimension of the features.
  • The researchers identified at-risk students using the proposed SLnSO-DQN model.

Sources:

  • Leveraging Hybrid Deep Q-Learning for Early Identification of At-Risk Students. Emerging Science Journal, 2025, 9():95-111.
  • Modern College of Business and Science, Social Science.
  • Ital Publication.
  • https://doi-org.sdpl.idm.oclc.org/10.28991/ESJ-2025-SIED1-06
  • P. Vijaya, Department of Mathematics and Computer Science, Modern College of Business and Science, Bowshar, Muscat.

Read more