Compression of Optimal Value Functions for Markov Decision Processes.
Mykel J. KochenderferNicholas MonathPublished in: DCC (2013)
Keyphrases
- markov decision processes
- dynamic programming
- average cost
- finite horizon
- average reward
- optimal policy
- state space
- action sets
- finite state
- transition matrices
- policy iteration
- stationary policies
- reinforcement learning
- decision theoretic planning
- infinite horizon
- reachability analysis
- discounted reward
- partially observable
- total reward
- model based reinforcement learning
- decision processes
- state and action spaces
- reinforcement learning algorithms
- risk sensitive
- factored mdps
- semi markov decision processes
- decision diagrams
- markov decision process
- image compression
- optimal solution
- planning under uncertainty
- action space
- reward function
- optimal control
- compressed data
- long run
- data compression
- compression ratio
- compression algorithm
- sufficient conditions
- search space