Duality and Policy Evaluation in Distributionally Robust Bayesian Diffusion Control

Jose Blanchet, Jiayi Cheng, Hao Liu, Yang Liu · 2025

Read Duality and Policy Evaluation in Distributionally Robust Bayesian Diffusion Control by Jose Blanchet, Jiayi Cheng, Hao Liu, Yang Liu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-19294.txt.

Introduction

Controllers that steer a system through noise often rely on a model whose parameters are only guessed. Jose Blanchet and co-authors build a safer scheme in a 2025 paper on distributionally robust Bayesian diffusion control. A Bayesian controller trusts its prior beliefs about the parameters, but a wrong prior can mislead it, while classic robust control can be too gloomy.

Their compromise lets an adversary nudge the prior, but only within a measured distance of the starting beliefs. The authors prove a strong duality result that collapses the robust evaluation into a small optimization problem. That yields a practical simulation method for scoring and learning policies. Tests on a synthetic control problem and on real portfolio data check that the machinery runs efficiently.

Catalog record

CREATORJose Blanchet, Jiayi Cheng, Hao Liu, Yang Liu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE159 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-19294

Browse the full library · About the Tower of Babel