Reinforcement learning is based on reinforcement theory in which a combination of reward and/or punishment is used to reinforce the wanted behavior or extinguish unwanted behavior. Software agents interact with the environment. In reinforcement learning agents can automatically figure out how to optimize their behavior given a system of rewards and punishment. It draws inspiration from behavioral psychology. It has applications in many fields. Such as:

• Economics
• Genetics
• Game Playing

In 2016, reinforcement learning was used to train google deep mind AlphaGO. AlphaGo is the first computer to defeat the world campaigns in GO.