Deepening the m-step Bellman lookahead in PGTS monotonically reduces the set of stationary policies, so the worst local optimum improves with depth.
Policy gradient with tree search (PGTS) in reinforcement learning evades local maxima
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Policy Gradient with Tree Search: Avoiding Local Optimas through Lookahead
Deepening the m-step Bellman lookahead in PGTS monotonically reduces the set of stationary policies, so the worst local optimum improves with depth.