Extending the Capabilities of Reinforcement Learning Through Curriculum: A Review of Methods and Applications

Reinforcement learning has long been advertised as the one with the capability to intelligently mimic and understand human learning and behavior. While the upshot of the field’s advances is not underrated, its applicability and extension to large, complex and highly dynamic environments remain inefficient, inaccurate or unsolved. Curriculum learning presents an intuitive yet elegant solution to these problems and when incorporated into the solution, provides a structured approach to alleviate some of the core challenges. As reinforcement learning framework proceeds to tackle harder challenges, it necessitates the study of essential support frameworks including curriculum learning. Through this paper, we review the current state-of-the-art in the field of curriculum-based reinforcement learning. We analyze and classify numerous scientific articles and present a summary of their methodologies and applications. In addition to the detailed review and analysis of the targeted algorithms, we summarise the current progress in the field by tabulating distinct identifying features of reviewed works with respect to their curriculum design methodology and applications. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd 2021.