Mixing Probabilistic and non-Probabilistic Objectives in Markov Decision Processes.
Raphaël BerthonShibashis GuhaJean-François RaskinPublished in: LICS (2020)
Keyphrases
- markov decision processes
- probabilistic planning
- dynamic programming
- probabilistic model
- factored mdps
- state space
- optimal policy
- reinforcement learning
- generative model
- infinite horizon
- finite state
- risk sensitive
- machine learning
- average cost
- planning under uncertainty
- partially observable
- multi agent
- model based reinforcement learning
- transition matrices