MBRL — Model-Based Reinforcement Learning

Fact

A reinforcement learning method using a learned world model to choose actions.

In Plain Words

MBRL is like trying a skateboard trick in a video game first. Your knees send a thank-you card.

It learns a mini world, then tries choices there before acting. Robots and self-driving cars use it to cut real test runs, but a bad model can steer them wrong.

Related Concepts

• Reinforcement Learning
• World model
• Markov Decision Process
• Monte Carlo Tree Search

RL
MBRL belongs to RL and learns the environment first.

World model
A World model acts like its practice world for predictions.

MDP
MBRL often uses an MDP to plan from state changes.

MCTS
MCTS can search a few moves inside its practice world.