Scholarlink

This study aims to further advance the existing research on adaptive learning by incorporating deep learning models and dataset refinements working toward enhancing performance. Prior research has demonstrated the effectiveness of various machine learning models like logistic regression and random forests which were able to achieve accuracies ranging from 66 to 92%. However, there was too much of variance in the accuracies and the models failed to generalize on different datasets. Therefore, though the accuracy was high, the models could not be trusted for their predictions. To remedy this, new models including simple artificial neural networks (ANN) and recurrent neural networks (RNN), especially bidirectional long short-term memory networks (Bi-LSTMs) were included. The results demonstrated that the Bi-LSTM model achieved a higher accuracy of around 85% as compared to the simple ANN models and other traditional machine learning methods. These developments highlight how machine learning is transforming teaching strategies (Gorski et al. in Educ Sci 13:1216, 2023; Zhu et al. in Sustainability 15:3115, 2023).

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing Adaptive Learning with Deep Learning Models: A Comparative Study of ANN and LSTM Approaches

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Junaid Mulla,
  • Snehal Rathi,
  • Vijayshri Khedkar,
  • Sharvil Nichat,
  • Sumit Mule,
  • Suyash Neware

摘要

This study aims to further advance the existing research on adaptive learning by incorporating deep learning models and dataset refinements working toward enhancing performance. Prior research has demonstrated the effectiveness of various machine learning models like logistic regression and random forests which were able to achieve accuracies ranging from 66 to 92%. However, there was too much of variance in the accuracies and the models failed to generalize on different datasets. Therefore, though the accuracy was high, the models could not be trusted for their predictions. To remedy this, new models including simple artificial neural networks (ANN) and recurrent neural networks (RNN), especially bidirectional long short-term memory networks (Bi-LSTMs) were included. The results demonstrated that the Bi-LSTM model achieved a higher accuracy of around 85% as compared to the simple ANN models and other traditional machine learning methods. These developments highlight how machine learning is transforming teaching strategies (Gorski et al. in Educ Sci 13:1216, 2023; Zhu et al. in Sustainability 15:3115, 2023).