Reinforcement Learning with Deep Energy-Based Policies.
Tuomas HaarnojaHaoran TangPieter AbbeelSergey LevinePublished in: ICML (2017)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- markov decision process
- control policies
- markov decision processes
- fitted q iteration
- state space
- reinforcement learning agents
- total reward
- reward function
- cooperative multi agent systems
- dynamic programming
- hierarchical reinforcement learning
- policy gradient methods
- partially observable markov decision processes
- function approximation
- markov decision problems
- control policy
- decision problems
- learning algorithm
- long run
- reinforcement learning algorithms
- model free
- infinite horizon
- temporal difference learning
- continuous state
- state and action spaces
- macro actions
- multi agent
- action space
- deep learning
- policy iteration
- decentralized control
- monte carlo
- multiagent reinforcement learning
- supervised learning
- optimal control