Evolution of Policy Gradient Methods in Reinforcement Learning: From Reinforce to EVM

Web: https://towardsdatascience.com/evolution-of-policy-gradient-methods-in-reinforcement-learning-from-reinforce-to-evm-4e80017a297b?source=rss----7f60cf5620c9---4

Evolution of Policy Gradient Methods in Reinforcement Learning

From Reinforce to EVM

Image by Author. Three robots represent three algorithms that are observed in this article: Reinforce, A2C and EVM (Empirical variance minimization)

Reinforcement learning (RL) is an area of machine learning where agents have to learn what actions to take interacting with an environment in order to maximize the cumulative reward. The basic setting is the following: an agent observes the state of the environment, chooses what action to …

evolution gradient learning machine learning neural networks policy reinforce reinforcement reinforcement learning