Optimization-Driven Adaptive Experimentation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Che, Ethan, Jiang, Daniel R., Namkoong, Hongseok, Wang, Jimmy
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910688958480384
author Che, Ethan
Jiang, Daniel R.
Namkoong, Hongseok
Wang, Jimmy
author_facet Che, Ethan
Jiang, Daniel R.
Namkoong, Hongseok
Wang, Jimmy
contents Real-world experiments involve batched & delayed feedback, non-stationarity, multiple objectives & constraints, and (often some) personalization. Tailoring adaptive methods to address these challenges on a per-problem basis is infeasible, and static designs remain the de facto standard. Focusing on short-horizon ($\le 10$) adaptive experiments, we move away from bespoke algorithms and present a mathematical programming formulation that can flexibly incorporate a wide range of objectives, constraints, and statistical procedures. We formulating a dynamic program based on central limit approximations, which enables the use of scalable optimization methods based on auto-differentiation and GPU parallelization. To evaluate our framework, we implement a simple heuristic planning method ("solver") and benchmark it across hundreds of problem instances involving non-stationarity, personalization, and multiple objectives & constraints. Unlike bespoke methods (e.g., Thompson sampling variants), our mathematical programming framework provides consistent gains over static randomized control trials and exhibits robust performance across problem instances.
format Preprint
id arxiv_https___arxiv_org_abs_2408_04570
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Optimization-Driven Adaptive Experimentation
Che, Ethan
Jiang, Daniel R.
Namkoong, Hongseok
Wang, Jimmy
Machine Learning
Real-world experiments involve batched & delayed feedback, non-stationarity, multiple objectives & constraints, and (often some) personalization. Tailoring adaptive methods to address these challenges on a per-problem basis is infeasible, and static designs remain the de facto standard. Focusing on short-horizon ($\le 10$) adaptive experiments, we move away from bespoke algorithms and present a mathematical programming formulation that can flexibly incorporate a wide range of objectives, constraints, and statistical procedures. We formulating a dynamic program based on central limit approximations, which enables the use of scalable optimization methods based on auto-differentiation and GPU parallelization. To evaluate our framework, we implement a simple heuristic planning method ("solver") and benchmark it across hundreds of problem instances involving non-stationarity, personalization, and multiple objectives & constraints. Unlike bespoke methods (e.g., Thompson sampling variants), our mathematical programming framework provides consistent gains over static randomized control trials and exhibits robust performance across problem instances.
title Optimization-Driven Adaptive Experimentation
topic Machine Learning
url https://arxiv.org/abs/2408.04570