Market-Based Reinforcement Learning in Partially Observable Worlds.
Ivo KweeMarcus HutterJürgen SchmidhuberPublished in: ICANN (2001)
Keyphrases
- partially observable
- reinforcement learning
- markov decision processes
- state space
- partial observability
- partially observable domains
- decision problems
- dynamical systems
- hidden state
- partially observable environments
- markov decision problems
- function approximation
- reward function
- infinite horizon
- action models
- partial observations
- belief state
- model free
- belief space
- multi agent
- temporal difference
- optimal policy
- learning algorithm
- reinforcement learning algorithms
- policy iteration
- partially observable markov decision processes
- optimal control
- dynamic programming
- partially observable markov decision process
- heuristic search
- search space
- action selection
- continuous state
- fully observable
- markov chain
- orders of magnitude