深度(层)学习:内涵、流变与展望

Deep Learning: Its Connotation, Evolution and Prospect

“Deep learning”同时存在于教育领域与计算机领域,二者的内涵有着差异,英文表述却相同。教育领域的“Deep learning”可以理解为“深层学习”,它是一种学习方法,同“教”与“学”活动的各个要素相关联。而计算机领域的“Deep learning”可以翻译为“深度学习”,它是作为人工智能的新兴方向,“特征学习”是其主要特征。国外教育领域深层学习的研究自1976年起,侧重于对影响学习者深层学习因素的探讨;国内的研究自2004年起,经历了概念明晰、模式应用到结果追寻等阶段。梳理分析国内外深度(层)学习的内涵和研究历程,可以化解“深度(层)学习”理解的概念迷思;对已有研究的批判性思考可以进一步地展望未来:即深度(层)学习将在横向上呈现“跨界融合”,在纵向上表现出理念深化、实践深入等研究趋势。

Deep learning is the same English wording shared by two different notions whose connotations are different in the field of education studies and computer science. Deep learning in the field of education can be understood as deep-level learning. It is a learning method that is related to the elements of teaching and learning activities. Deep learning in the computer science can be translated as shendu xuexi 深度学习 in Chinese and is an emerging direction of the development of artificial intelligence,which is mainly characterized by feature learning. Researches on deep learning abroad have focused on the issue of what factors influence learners’ deep-level learning since 1976. In China,since 2004 researches in this field have gone through the following three stages: the deliberation on the concept;the application of the teaching mode it advocates;and the pursuit of results. Sorting out and analyzing the connotation and research process of deep learning at home and abroad can resolve the misunderstandings of deep learning. A critical analysis of the previous studies on deep learning can help us better understand the direction in the future: that is,the research on deep learning will present a trend of interdisciplinary integration and the deepening of both ideas and practice.