Logarithmic Smoothing for Pessimistic Off-Policy Evaluation, Selection and Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sakhi, Otmane, Aouali, Imad, Alquier, Pierre, Chopin, Nicolas
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916461644087296
author Sakhi, Otmane
Aouali, Imad
Alquier, Pierre
Chopin, Nicolas
author_facet Sakhi, Otmane
Aouali, Imad
Alquier, Pierre
Chopin, Nicolas
contents This work investigates the offline formulation of the contextual bandit problem, where the goal is to leverage past interactions collected under a behavior policy to evaluate, select, and learn new, potentially better-performing, policies. Motivated by critical applications, we move beyond point estimators. Instead, we adopt the principle of pessimism where we construct upper bounds that assess a policy's worst-case performance, enabling us to confidently select and learn improved policies. Precisely, we introduce novel, fully empirical concentration bounds for a broad class of importance weighting risk estimators. These bounds are general enough to cover most existing estimators and pave the way for the development of new ones. In particular, our pursuit of the tightest bound within this class motivates a novel estimator (LS), that logarithmically smooths large importance weights. The bound for LS is provably tighter than its competitors, and naturally results in improved policy selection and learning strategies. Extensive policy evaluation, selection, and learning experiments highlight the versatility and favorable performance of LS.
format Preprint
id arxiv_https___arxiv_org_abs_2405_14335
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Logarithmic Smoothing for Pessimistic Off-Policy Evaluation, Selection and Learning
Sakhi, Otmane
Aouali, Imad
Alquier, Pierre
Chopin, Nicolas
Machine Learning
This work investigates the offline formulation of the contextual bandit problem, where the goal is to leverage past interactions collected under a behavior policy to evaluate, select, and learn new, potentially better-performing, policies. Motivated by critical applications, we move beyond point estimators. Instead, we adopt the principle of pessimism where we construct upper bounds that assess a policy's worst-case performance, enabling us to confidently select and learn improved policies. Precisely, we introduce novel, fully empirical concentration bounds for a broad class of importance weighting risk estimators. These bounds are general enough to cover most existing estimators and pave the way for the development of new ones. In particular, our pursuit of the tightest bound within this class motivates a novel estimator (LS), that logarithmically smooths large importance weights. The bound for LS is provably tighter than its competitors, and naturally results in improved policy selection and learning strategies. Extensive policy evaluation, selection, and learning experiments highlight the versatility and favorable performance of LS.
title Logarithmic Smoothing for Pessimistic Off-Policy Evaluation, Selection and Learning
topic Machine Learning
url https://arxiv.org/abs/2405.14335