Reward Shaping

Fact

A way to guide an agent by changing the rewards it gets while learning.

In Plain Words

Reward Shaping is like giving a kid gold stars for each math step. You do not wait for the final answer to throw a parade.

It helps RL learn faster. You see it in robots and game AI.

Related Concepts

• Reinforcement Learning
• Markov Decision Process
• Reward Hacking

RL
Reward Shaping changes rewards so RL can learn a good policy faster.

MDP
Reward Shaping adjusts rewards in an MDP, but should not change the final goal.

Reward Hacking
If the reward is poorly shaped, the agent may learn to game the rules.