Deep Active Inference for Partially Observable MDPs.
Otto van der HimstPablo LanillosPublished in: IWAI (2020)
Keyphrases
- partially observable
- markov decision processes
- reinforcement learning
- markov decision problems
- state space
- decision problems
- dynamical systems
- infinite horizon
- partial observability
- reward function
- partial observations
- action models
- partially observable environments
- probabilistic inference
- belief state
- finite state
- optimal policy
- probabilistic planning
- partially observable markov decision process
- fully observable
- decision making
- reinforcement learning algorithms
- markov decision process
- policy iteration
- sufficient conditions
- probabilistic model
- search algorithm