In CRMDPs with spiky reward corruption, the environment is solvable; an algorithm detects corrupt states, fully characterizes the regret bound, and enables optimal policy learning with common RL algorithms.
Taylor, Brian Kulis, and Fei Sha
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Detecting Spiky Corruption in Markov Decision Processes
In CRMDPs with spiky reward corruption, the environment is solvable; an algorithm detects corrupt states, fully characterizes the regret bound, and enables optimal policy learning with common RL algorithms.