Introduces an L*-based active learning algorithm for deterministic MDPs that uses trace sampling to infer complete models including states and outperforms passive methods on accuracy with equal data.
Title resolution pending
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
-
L*-Based Learning of Markov Decision Processes (Extended Version)
Introduces an L*-based active learning algorithm for deterministic MDPs that uses trace sampling to infer complete models including states and outperforms passive methods on accuracy with equal data.