Model-Based vs Model-Free RL

Definition

A reinforcement-learning distinction: model-free RL directly learns action values from experience over a finite action space (e.g. basal ganglia, evolution, culture), while model-based RL uses a learned world model to predict consequences and rewards and to plan (e.g. cortex). The brain layers both.

Cast within

No topic-region cast yet — this appears once Model-Based vs Model-Free RL's compiled claims are aligned into a topic region's argument tree.

Claims mentioning Model-Based vs Model-Free RL (3)

Your notes

Loading notes...