Detecting Spiky Corruption in Markov Decision Processes.
Jason MancusoTomasz KisielewskiDavid LindnerAlok SinghPublished in: CoRR (2019)
Keyphrases
- markov decision processes
- finite state
- optimal policy
- transition matrices
- state space
- reinforcement learning
- policy iteration
- decision theoretic planning
- factored mdps
- dynamic programming
- finite horizon
- planning under uncertainty
- reinforcement learning algorithms
- partially observable
- infinite horizon
- reachability analysis
- decision processes
- risk sensitive
- average reward
- average cost
- state and action spaces
- data mining
- state abstraction
- action sets
- reward function