Reinforcement Learning Research: A Scientometric Assessment of Global publications output during 2009-18

Abstract

The paper provides quantitative and qualitative description of reinforcement learning research based on global publications output (5345 publications) in the field as indexed in Scopus database during 2009-16. Reinforcement learning research registered 11.5% growth per annum and averaged 17.11 citations per paper in 10 years. Top 10 most productive countries account for 97.17% global publication share.  The paper characterizes performance of reinforcement learning research on measures such as relative citation index, citations per paper, highly cited papers, top 20 global organizations and top 20 global authors in the field. The study identifies hot areas of research in the subject.   The study concludes that the USA, China, the UK, Japan, and Germany are the mainstream nations in reinforcement learning research whereas rest of the world countries is peripheral to research in the subject. The challenge is how to bring other world leading economies like India, Russia and France to the mainstream research in the subject.