A Gentle Introduction to Reinforcement Learning (RL)
Introduction to Reinforcement Learning (RL)
RL is a subfield of Artificial Intelligence (AI) that allows an agent to learn how to achieve a desired outcome, or maximize rewards over time, through trial and error learning. In this post, I give an introduction to the basics of RL, including some common terminology and example problems. I will also discuss some of the advantages and drawbacks of RL compared to other popular machine learning algorithms. Stay tuned for future posts in which I will go into more detail on specific RL techniques!
What is RL and Why Should You Care About it?
RL is a subfield of AI that deals with how agents should select controls in an environment so as to maximize some notion of cumulative reward. RL is considered by many to be the third generation of machine learning algorithms, after supervised learning (SL) and unsupervised learning (UL).
The notation used in RL is similar to that used in SL, but with two important differences. First, in RL there is no training data per se (labeled data); instead the agent must learn from experience by trial and error (hence the “reinforcement” part of the name). Second, most RL problems are formalized as Markov decision processes (MDPs), which are a specific…
