Reinforcement Learning - A Systematic Literature Review

Abstract

Supervised Learning, Unsupervised Learning, and Reinforcement Learning (RL) are the three main categories of Machinelearning (ML). Supervised learning involves pre-training a model on a labeled dataset. On the other hand, in unsupervisedlearning, the model is trained on unlabeled data.In contrast, RL is motivated by evaluating feedback rather than labels. Here, the agent learns the optimal pathfor solving sequential decision-making problems by interacting with the environment and taking the most suitable course ofaction in a given situation in order to maximize the reward. The RL agent makes its own decisions on how to complete atask. Additionally, there is no training data; as a result, the agent learns by accumulating experience. RL has applications ina wide variety of fields, including natural sciences, transportation, finance, and engineering, as well as smart grids, robotics,and healthcare.