Towards a Learning Style and Knowledge Level-Based Adaptive Personalized Platform for an Effective and Advanced Learning for School Students
Towards a Learning Style and Knowledge Level-Based Adaptive Personalized Platform for an Effective and Advanced Learning for School Students
复制标题
建立基于学习方式和知识水平的自适应个性化平台，为中学生提供有效和先进的学习
DOI：
--
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Samah El
中科院分区：
文献类型：
--
作者：
W. Sayed;M. Gamal;Moemen Abdelrazek;Samah El
本章提出了一种基于人工智能（AI）的自适应个性化平台，以实现有效和高级的学习。大多数电子学习平台都针对成人和终身学习者。然而，通过电子学习和人工智能 (AI) 支持，可以大大增强年轻人的教育过程，以适应每个学习者的节奏和学习风格。此外，它还补充了课堂教师的角色，为每个学习者提供与他/她的能力、偏好和需求相匹配的一对一辅导。基于问题的数学定义，我们发现强化学习（RL）是最适合所提出的针对学生的自适应个性化电子学习系统的人工智能技术。对相关研究工作进行了文献综述，一方面关注学校学生的个性化电子学习系统，另一方面利用强化学习解决这个问题。在提议的系统设计中考虑了学习风格、视觉、听觉、读/写和动觉（VARK）以及布鲁姆分类法。设计了一个基于Moodle学习管理系统(LMS)的网站作为电子学习平台。负责适应的人工智能模块（AIM）是使用多任务深度 Q 学习开发的。该模块是使用电子贪婪策略来实现和训练的。它的性能是使用奖励函数的运行平均值、总分类损失和 VARK 损失来评估的。性能指标验证了 RL 算法的收敛性。
This chapter proposes an artificial intelligence (AI)-based adaptive personalized platform for an effective and advanced learning. Most e-learning platforms target adult and lifelong learners. Yet, the educational process for younger people can be much enhanced through e-learning and artificial intelligence (AI) support to suit each learner’s pace and learning style. In addition, it complements the role of classroom teacher in providing one-to-one tutoring for each learner, which is matched to his/her capabilities, preferences, and needs. Based on the mathematical definition of the problem, it is found that reinforcement learning (RL) is the most suitable AI technique for the proposed adaptive personalized e-learning system for school students. A literature review of the related research works is provided focusing on personalized e-learning systems for school students on the one hand and utilizing RL in this problem on the other hand. Learning styles, visual, aural, read/write, and kinesthetic (VARK), and Bloom’s taxonomy are considered in the proposed system design. A website is designed based on Moodle learning management system (LMS) as the e-learning platform. An artificial intelligence module (AIM) responsible for adaptation is developed using multitask deep Q-learning. The module is implemented and trained using an e-greedy policy. Its performance is evaluated using the running mean of the reward function, the total taxonomy loss, and the VARK loss. The performance metrics validate the convergence of the RL algorithm.