Gnu-RL: A Precocial Reinforcement Learning Solution for Building HVAC Control Using a Differentiable MPC Policy.
Bingqing ChenZicheng CaiMario BergésPublished in: BuildSys@SenSys (2019)
Keyphrases
- control policy
- reinforcement learning
- optimal policy
- control policies
- optimal control
- action selection
- control problems
- approximate dynamic programming
- rl algorithms
- policy search
- long run
- actor critic
- markov decision processes
- policy evaluation
- average cost
- semi markov decision process
- reinforcement learning algorithms
- partially observable domains
- state space
- adaptive control
- model free
- function approximation
- policy iteration
- action space
- policy gradient
- reward function
- continuous state
- markov decision process
- learning algorithm
- feedback control
- control strategies
- partially observable
- stochastic control
- markov decision problems
- temporal difference
- mathematical model
- machine learning
- partially observable environments
- state and action spaces
- objective function
- open source
- transfer learning
- learning problems
- reinforcement learning methods
- state action
- dynamic model
- function approximators
- infinite horizon