Regret Minimization in MDPs with Options without Prior Knowledge.
Ronan FruitMatteo PirottaAlessandro LazaricEmma BrunskillPublished in: NIPS (2017)
Keyphrases
- regret minimization
- prior knowledge
- markov decision processes
- nash equilibrium
- game theoretic
- reinforcement learning
- factored mdps
- state space
- game theory
- optimal policy
- finite horizon
- decision theoretic planning
- domain knowledge
- training data
- partially observable
- semi markov decision processes
- probabilistic planning
- real time dynamic programming
- factored markov decision processes
- markov decision problems
- policy iteration
- markov decision process
- finite state
- decision problems
- decision diagrams
- option pricing
- planning under uncertainty
- initial state