Learning is driven by reward

What will you get in this article

  • Basic concept of reinforcement learning

Outline

Concept

In many scenario, real world problems do not exist the ground true answer. The supervised learning could not handle this problem.

But we can use trial and error to get the answer then calibrate our behavior. The key concept is reward from environment. That is also the way human learning new thing.

Example

Key Components of Reinforce Learning (RL)

  • Observation
  • Action
  • Reward

Goal: Maximize total future reward ( return )