Mapping the Research Landscape of Deep Learning from 2001 to 2019
Fatih Gürcan, Setenay Şevik · 2019 1st International Informatics and Software Engineering Conference (UBMYK) · 2019
This paper analyzes recent literature to investigate the emerging topics and trends in deep learning research. In this context, a semantic content analysis based on probabilistic topic modeling has been performed on 22,279 journal articles on the subject of deep learning during the last 19 years between 2001 and 2019. As a consequence of this analysis, 18 topics mapping the research landscape of deep learning was revealed. The topic "Classification" has the highest ratio (10.57%). It is followed by the topics "Image feature representing" (8.90%), and "Data applications" (8.71%). In addition, with the purpose of identifying the time-dependent average and acceleration of these topics, the average number of articles for each topic, and the rate of increase in the number of articles for each topic were calculated. In this context, the topics were separated into three groups including "hot topics", "stable topics", and "cold topics". These discovered topics, which map the research landscape of deep learning, are expected to contribute to understanding the present and future of deep learning.