Softmax policy mirror ascent is a normalization-free mirror ascent on logits that converges linearly in tabular MDPs and linearly to a neighborhood with function approximation.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Fast Convergence of Softmax Policy Mirror Ascent
Softmax policy mirror ascent is a normalization-free mirror ascent on logits that converges linearly in tabular MDPs and linearly to a neighborhood with function approximation.