Sampling Humans for Optimizing Preferences in Coloring Artwork

TL;DR

This paper reviews a Bayesian optimization method for preference-based optimization, adapts it to handle ties, and discusses challenges faced when humans participate in optimizing artwork coloring.

Contribution

It introduces an adaptation of Bayesian optimization for preferences to include ties and explores human-in-the-loop challenges in artwork coloring tasks.

Findings

Bayesian optimization can be adapted to handle ties in preferences.

Humans encounter specific difficulties when using preference optimization for artwork.

Insights into human-in-the-loop optimization challenges are discussed.

Abstract

Many circumstances of practical importance have performance or success metrics which exist implicitly---in the eye of the beholder, so to speak. Tuning aspects of such problems requires working without defined metrics and only considering pairwise comparisons or rankings. In this paper, we review an existing Bayesian optimization strategy for determining most-preferred outcomes, and identify an adaptation to allow it to handle ties. We then discuss some of the issues we have encountered when humans use this optimization strategy to optimize coloring a piece of abstract artwork. We hope that, by participating in this workshop, we can learn how other researchers encounter difficulties unique to working with humans in the loop.