【作 者】：
李 默 妍
【关 键 词 】：
联邦学习； 教育大数据； 教育数据挖掘； 隐私保护； 机器学习
【栏 目】：
学习环境与资源
【中文摘要】：
近年来，人工智能在教育领域发挥着日益重要的作用。但随着隐私泄露问题的凸显，如何在保护学习者隐私的基础上，使用来自多方的数据以提升人工智能应用的性能，成为智能时代亟待解决的问题。为此，文章引入了人工智能领域新兴的联邦学习概念，分析了联邦学习的定义、系统模型与训练过程、隐私保护技术，并将联邦学习与教育数据挖掘的各类算法相结合，以解决教育数据挖掘中可能存在的隐私保护问题。研究发现，联邦学习方法能够从原理上保障数据隐私，且容易整合到现有的教育应用中；在保护隐私的基础之上，运用联邦学习能够最大程度地提高模型精确度；将联邦学习与教育数据挖掘相结合，既能最大化地发挥利益相关者的作用，又能满足各利益相关者的需求。联邦学习将为教育的信息化与智能化发展开辟全新的路径。
【英文摘要】：
In recent years, artificial intelligence has played an increasingly important role in education. However, with the prominence of privacy disclosure , how to use data from various sources to improve the performance of AI applications while protecting the privacy of learners has become an urgent issue in the intelligent era. Therefore, this paper introduces the emerging concept of federated learning in the field of artificial intelligence, analyzes the definition of federated learning, system model and training process and privacy protection techniques, and combines federated learning with various algorithms for educational data mining in order to solve the possible privacy protection problems in educational data mining. It is found that the federated learning approach can guarantee data privacy in principle and can be easily integrated into existing educational applications. The use of federated learning can maximize the accuracy of the model while protecting privacy. Combining federated learning with educational data mining will both maximize the role of stakeholders and meet the needs of each stakeholder. Federated learning will open up a new path for the development of informatization and intellectualization in education.