policy_gradient.md

tags	python,numpy,neural-network,reinforcement learning
mathjax	true

Policy Gradient Methods

• learns state to action mapping directly which is often more simple
• no model of environment dynamics needed
• allows continuous action space
• allows stochastic policy which can be a crucial advantage compared to deterministic policies
• Actor Critic RL Methods

{:.caption .img} Hado Van Hasselt - Reinforcement Learning, Policy Gradients and Actor Critics (2018)