None To Optima in Few Shots: Bayesian Optimization with MDP Priors

Diantong Li, Kyunghyun Cho, Chong Liu · 2025

Read None To Optima in Few Shots: Bayesian Optimization with MDP Priors by Diantong Li, Kyunghyun Cho, Chong 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-2511-01006.txt.

Description

Diantong Li, Kyunghyun Cho, and Chong Liu improve Bayesian optimization for problems where each test is expensive, such as drug discovery or materials design. Bayesian optimization is efficient, but its guarantees are mostly long run results, and real budgets may allow only a handful of evaluations.

Their algorithm, ProfBO, learns from how optimization went on related earlier tasks. Those past runs are modeled as Markov decision processes. They capture procedural know how about searching well. They are then embedded into a neural network that adapts quickly to a new task through meta learning. On COVID and cancer benchmarks and on hyperparameter tuning, ProfBO found high quality solutions with far fewer evaluations than state of the art methods.

Catalog record

CREATORDiantong Li, Kyunghyun Cho, Chong Liu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE47 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-01006

Browse the full library · About the Tower of Babel