Deep imagination is a close to optimal policy for planning in large decision trees under limited resources.
Rubén Moreno-BoteChiara MastrogiuseppePublished in: CoRR (2021)
Keyphrases
- limited resources
- optimal policy
- decision trees
- markov decision processes
- finite horizon
- infinite horizon
- decision problems
- processing power
- reinforcement learning
- initial state
- state space
- partially observable markov decision processes
- dynamic programming
- state dependent
- long run
- markov decision problems
- planning problems
- multistage
- finite state
- partially observable
- average cost
- control policies
- bayesian reinforcement learning
- serial inventory systems
- machine learning
- naive bayes
- production planning
- sufficient conditions
- policy iteration
- markov decision process
- stochastic demand
- periodic review
- expected reward
- cost function
- learning algorithm