The generalized linear MDP is Bellman complete, and pessimistic offline algorithms achieve parametric suboptimality rates using labeled data plus, in the semi-supervised variant, unlabeled transitions.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Generalized Linear Markov Decision Process
The generalized linear MDP is Bellman complete, and pessimistic offline algorithms achieve parametric suboptimality rates using labeled data plus, in the semi-supervised variant, unlabeled transitions.