Title of Abstract

The Effects of Multiobjective Task Performance on Heterogenous and Homogenous Multiagent Learning

Poster Number

61

Submitting Student(s)

Rachel Moan

Faculty Sponsor (for work done with a non-Winthrop mentor)

Paul Wiegand, Ph.D.; Kristen Abernathy, Ph.D.; Trent Kull, Ph.D.; & Michael Lipscomb, Ph.D.

College

College of Business Administration

Department

Computer Science & Quantitative Methods

Abstract

Cooperative team learning can be categorized as heterogeneous, where the agents learn specialized behaviors, or homogeneous, where agents learn the same behavior. Prior work has compared these two types of learning, with varying degrees of success in each case. What we lack, however, is an understanding of when heterogeneity should be used rather than homogeneous methods. The goal of this work is to understand how the circumstances of a problem affect the success of a team. Specifically, we explore how conflicting objectives affect the performance of heterogeneous and homogeneous teams. We examine this question by presenting multiple simulated scenarios, each with varying degrees of conflict in the team’s objectives, and compare the performance of homogeneous and heterogeneous teams of agents.

Start Date

15-4-2022 12:00 PM

This document is currently not available here.

Share

COinS
 
Apr 15th, 12:00 PM

The Effects of Multiobjective Task Performance on Heterogenous and Homogenous Multiagent Learning

Cooperative team learning can be categorized as heterogeneous, where the agents learn specialized behaviors, or homogeneous, where agents learn the same behavior. Prior work has compared these two types of learning, with varying degrees of success in each case. What we lack, however, is an understanding of when heterogeneity should be used rather than homogeneous methods. The goal of this work is to understand how the circumstances of a problem affect the success of a team. Specifically, we explore how conflicting objectives affect the performance of heterogeneous and homogeneous teams. We examine this question by presenting multiple simulated scenarios, each with varying degrees of conflict in the team’s objectives, and compare the performance of homogeneous and heterogeneous teams of agents.