Value-based vs Policy-based Reinforcement Learning

1. Policy-based Reinforcement Learning

Suppose we have a good policy (a|s).

Upon observing the stats ,

random sampling: ~(.|).

2. Policy-based Reinforcement Learning

Suppose we know the optimal action-value function .

Upon observe the state ,

choose the action that maximizes the value:   = .

3. Summary

The objective of Reinforcement Learning is to learn good policy (a|s) and/or optimal action-value function .