人工类别学习研究范式述评

A Review on Research Paradigm of Artificial Category Learning

人工类别学习的研究对于理解以概念/规则形成为中心的学习活动有重要意义,到目前为止,有代表性的人工类别学习研究范式主要有基于规则的任务、信息整合任务、原型变形任务和天气预报。我们对这些任务在人类学习问题研究中的价值和启示做了讨论,也提出类别归纳加工在理解类别学习中可能具有核心地位,这些分析将有益于开展更具整合性的类别学习研究。

The researches of artificial category learning play an important role in understanding human learning in which the concept or rule formation is central; the representative paradigms of artificial category learning were reviewed in this paper. So far, the prevailing tasks of category learning include the rule-based tasks, the information-integration tasks, the prototype distortion tasks and the weather prediction tasks. Their values and inspiration in human learning behaviors were discussed, and category induction was thought to play a central role in category learning. These analyses would benefit more integrative researches on category learning.