Discovering User Types: Mapping User Traits by Task-Specific Behaviors in Reinforcement Learning

TL;DR

This paper introduces a method to classify user types based on task-specific behaviors in reinforcement learning, enabling rapid personalization of interventions across similar environments by formalizing environment equivalence classes.

Contribution

It proposes a novel framework for mapping user traits to behaviors in RL and formalizes environment equivalence to transfer intervention strategies.

Findings

Different environments can share the same set of user types.

Environment equivalence classes enable transfer of intervention strategies.

The approach facilitates rapid personalization of user interventions.

Abstract

When assisting human users in reinforcement learning (RL), we can represent users as RL agents and study key parameters, called \emph{user traits}, to inform intervention design. We study the relationship between user behaviors (policy classes) and user traits. Given an environment, we introduce an intuitive tool for studying the breakdown of "user types": broad sets of traits that result in the same behavior. We show that seemingly different real-world environments admit the same set of user types and formalize this observation as an equivalence relation defined on environments. By transferring intervention design between environments within the same equivalence class, we can help rapidly personalize interventions.