Actor-Critic Methods

Actor-Critic Methods

• Actor-critic agent is an agent that uses function approximation to learn a policy and value function
• Employ two NN
  • One for Actor
  • One for Policy
• Advantages:
  • more stable than value-based agents
  • Need fewer samples than policy-based agents

Generalized Advantage Estimation (GAE)

DDPG

• Deep Deterministic Policy Gradient
• DDPG is a DQN method for continuous action spaces
• DDPG uses a replay buffer and soft updates to the target networks

Read

• Q-Prop: Sample-Efficient Policy Gradient with An Off-Policy Critic
• High-Dimensional Continuous Control Using Generalized Advantage Estimation

• ← Previous
Next →