自适应人工智能在线教育关键技术研究

Artificial Intelligence has become a critical technology all around the world. New technologies associated with artificial intelligence is shaping a new revolution of the industrial technology. However, the education of artificial intelligence has not been a specialist education in China yet and only a small number of people can receive systematic study for some specialized agencies. It therefore is not conducive to construction talent echelon of artificial intelligence and impeding the development of artificial intelligence in China. This proposal intends to overcome the difficulties encountered in the education of artificial intelligence from four aspects: (1) Integrating knowledge map and discovery technology to establish artificial intelligence knowledge base for the organic organization and knowledge storage; (2) Exploring simple expressions of professional knowledge to overcome the disadvantages of abstracting knowledge and difficulties in artificial intelligence training; (3) Studying the self-adaptive recommendation algorithm based on individual preference to address the deficiency of the existing indistinguishable teaching model in individualized teaching; (4) Developing an interactive verification platform with a feedback mechanism to solve the verification difficulties of artificial intelligence knowledge. The proposal aims at developing a suite of methods and tools for the education of artificial intelligence, providing new approaches for non-specialist education and universal education of artificial intelligence knowledge, contributing to promote the popularization of the artificial intelligence foundation education and the flattening of the barriers to entry. The research results have important scientific significance and social value.

人工智能已成为全球科技研究的一个制高点，以人工智能为代表的新技术带来了工业科技的又一次革命。然而，人工智能教育迄今为止在我国尚属专才教育，只有少部分人能进入专门机构进行系统学习，这种现状不利于我国人工智能人才梯队的建设，进而将制约我国人工智能乃至信息技术未来的发展与壮大。本项目拟从四个层面研究解决人工智能进入全民教育阶段面临的难点，包括：结合知识图谱和发现技术建立人工智能知识库，依据智能关联将知识进行重新组织与存储；探索深入浅出的专业知识可视化表达与传播，克服人工智能知识抽象、传授困难的弊端；研究基于个体差异的自适应推荐算法，弥补无差异教学模式在个性化教学上的不足；研究带反馈机制的交互式验证平台，解决人工智能知识验证难的问题。本项目提出的自适应人工智能教育方式，能为人工智能普及化教育及非专才教育提供新途径，有助于推进人工智能基础教育全民化、准入门槛平坦化，具有重要的科学意义和社会价值。

人工智能已成为全球科技研究的一个制高点，正在带来工业科技的又一次革命。然而人工智能在教育领域的应用仍面临许多困难。本项目从四个层面研究解决了人工智能进入课堂教育和在线教育面临的难点，包括：建立了基于多媒体教室和移动设备的交互式移动教辅系统，实现了细粒度的教学追踪和综合建模，实现了教学过程的数据化、抽象化、可视化，支持实时教学评估和智能规划；探索了教学可视化表达与传播，研究了带反馈机制的交互式验证平台，建立了智能移动教辅系统平台，解决了知识抽象、传授困难的弊端；研究了基于生成对抗网络的在线教育数据增强方法，将深度强化学习方法引入到模仿学习中，通过生成虚拟学生仿真教学过程，克服了教育数据稀缺和不平衡的弱点；研究了基于深度强化学习的多目标自适应教学推荐算法，弥补了无差异教学模式在个性化教学、多目标教学上的不足，实现了模型无关的、具有较高的普适性的多目标导向教学路径规划。本项目提出的自适应人工智能教育方式，从教学工具、教学数据、教学方法三个方向来支撑自适应人工智能在线教育关键技术研究，能为人工智能在课堂教育和在线教育方面提供新途径，有助于推进人工智能教育应用实用化、准入门槛平坦化，具有重要的科学意义和社会价值。