Computer Science and Engineering, Department of

 

First Advisor

Leen-Kiat Soh

Date of this Version

6-2017

Document Type

Article

Comments

A THESIS Presented to the Faculty of The Graduated College at the University of Nebraska In Partial Fulfillment of Requirements For the Degree of Master of Science, Major: Computer Science, Under the Supervision of Professor Leen-Kiat Soh. Lincoln, Nebraska: June, 2017

Copyright (c) 2017 Bin Chen.

Abstract

When deciding which ad hoc team to join, agents are often required to consider rewards from accomplishing tasks as well as potential benefits from learning when working with others, when solving tasks. We argue that, in order to decide when to learn or when to solve task, agents have to consider the existing agents’ capabilities and tasks available in the environment, and thus agents have to consider agent and task openness—the rate of new, previously unknown agents (and tasks) that are introduced into the environment. We further assume that agents evolve their capabilities intrinsically through learning by observation or learning by doing when working in a team. Thus, an agent will need to consider which task to do or which team to join would provide the best situation for such learning to occur. In this thesis, we develop an auction-based multiagent simulation framework, a mechanism to simulate openness in our environment, and conduct comprehensive experiments to investigate the impact of agent and task openness. We propose several agent task selection strategies to leverage the environmental openness. Furthermore, we present a multiagent solution for agent-based collaborative human task assignment when finding suitable tasks for users in complex environments is made especially challenging by agent openness and task openness. Using an auction-based protocol to fairly assign tasks, software agents model uncertainty in the outcomes of bids caused by openness, then acquire tasks for people that maximize both the user’s utility gain and learning opportunities for human users (who improve their abilities to accomplish future tasks through learning by experience and by observing more capable humans). Experimental results demonstrate the effects of agent and task openness on collaborative task assignment, the benefits of reasoning about openness, and the value of non-myopically choosing tasks to help people improve their abilities for uncertain future tasks.

Share

COinS