Learning Automata: A Comparative Analysis of Estimator Algorithms

Learning automata (LAs) play a crucial role as a reinforcement scheme to solve engineering problems even in nonstationary environments. However, their low rate of convergence has considered as the main drawback in the works of literature. Estimator algorithms have suggested as a successful attempt to alleviate the inconvenience. The aim of this paper is to discuss various types of estimator algorithms and categorize them based on their intrinsic properties. Also, an overall comparison between existing algorithms based on the conventional measures is presented. The associated analysis provides a foundation for identifying strengths and weaknesses of estimator algorithms in the field of LAs, as well as general guidelines for future improvements and innovations. Keywords-Bayesian estimator; estimator algorithms; learning automata; maximum likelihood estimator.