Hindsight is Only 50/50: Unsuitability of MDP based Approximate POMDP Solvers for Multi-resolution Information Gathering.
Sankalp AroraSanjiban ChoudhurySebastian A. SchererPublished in: CoRR (2018)
Keyphrases
- information gathering
- multiresolution
- markov decision processes
- markov decision process
- optimal policy
- finite state
- reinforcement learning
- policy evaluation
- state space
- partially observable markov decision processes
- partially observable
- resource bounded
- information fusion
- partially observable markov decision process
- markov decision problems
- planning under uncertainty
- reward function
- policy iteration
- point based value iteration
- decision making
- bayesian reinforcement learning
- state and action spaces
- continuous state
- dynamic programming
- decision process
- decision problems
- infinite horizon
- dynamical systems
- sat solvers
- artificial intelligence
- average reward
- linear programming
- decision support
- data fusion
- average cost
- utility function
- belief state
- heuristic search
- approximate solutions
- computational intelligence
- artificial neural networks
- multi agent systems
- decision theoretic
- dec pomdps
- decision theoretic planning
- soft computing
- learning algorithm
- machine learning