TensorPlan and the Few Actions Lower Bound for Planning in MDPs under Linear Realizability of Optimal Value Functions.
Gellért WeiszCsaba SzepesváriAndrás GyörgyPublished in: CoRR (2021)
Keyphrases
- lower bound
- stochastic domains
- optimal solution
- worst case
- upper bound
- partially observable
- markov decision processes
- action sets
- decision theoretic planning
- initial state
- state space
- goal state
- plan recognition
- macro actions
- decision theoretic
- reinforcement learning
- average cost
- planning problems
- closed form
- planning under uncertainty
- competitive ratio
- linear functions
- action selection
- optimal policy
- dynamic programming
- objective function
- branch and bound
- lower and upper bounds
- probabilistic planning
- heuristic search
- domain independent
- optimal plans
- situation calculus
- regret bounds
- ai planning
- infinite horizon
- planning graph
- optimal planning
- finite horizon
- markov decision problems
- constant factor
- optimal strategy
- decision processes
- decision problems
- external events
- classical planning