AWhat changed
Used an expectation-maximization style policy-improvement step with constrained nonparametric targets followed by parametric policy fitting; V-MPO adapts this approach to on-policy, value-based advantage weighting.
Scientific scope
Used an expectation-maximization style policy-improvement step with constrained nonparametric targets followed by parametric policy fitting; V-MPO adapts this approach to on-policy, value-based advantage weighting. Bundled off-policy MPO2018 and on-policy V-MPO2019 with distinct original bylines, not one common author list. Inherits EM, relative-entropy policy search and policy iteration; KL constraints and reported stability do not ensure globally optimal neural policies or arbitrary trust-region equivalence.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Bundled off-policy MPO2018 and on-policy V-MPO2019 with distinct original bylines, not one common author list. Inherits EM, relative-entropy policy search and policy iteration; KL constraints and reported stability do not ensure globally optimal neural policies or arbitrary trust-region equivalence.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
- Maximum a Posteriori Policy Optimisation
- V-MPO: On-Policy Maximum a Posteriori Policy Optimization for Discrete and Continuous Control
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction