國立虎尾科技大學 |

Bounty Hunting : = A Dynamic Multiagent Task Allocation Mechanism.

紀錄類型:	書目-語言資料,手稿 : Monograph/item
正題名/作者:	Bounty Hunting :/
其他題名:	A Dynamic Multiagent Task Allocation Mechanism.
作者:	Wicke, Drew.
面頁冊數:	1 online resource (135 pages)
附註:	Source: Dissertation Abstracts International, Volume: 79-11(E), Section: B.
Contained By:	Dissertation Abstracts International79-11B(E).
標題:	Computer science. -
電子資源:	click for full text (PQDT)
ISBN:	9780438115194

Bounty Hunting : = A Dynamic Multiagent Task Allocation Mechanism.
Wicke, Drew.

Bounty Hunting :A Dynamic Multiagent Task Allocation Mechanism. - 1 online resource (135 pages)

Source: Dissertation Abstracts International, Volume: 79-11(E), Section: B.

Thesis (Ph.D.)--George Mason University, 2018.

Includes bibliographical references

Fundamental to many applications of multiagent systems is the problem of dynamically allocating tasks to agents that can solve them. Such dynamic task-allocation problems occur in a variety of real-life scenarios, ranging from automated delivery services to disaster response. The most popular current approach to solving this problem is based on auctioning tasks to the agents. This relies on nonintuitive assumptions such as an infinite money supply, exclusive allocation of tasks, and lack of incentive to complete the task. Such assumptions fail to hold in many real-world scenarios.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2018

Mode of access: World Wide Web

ISBN: 9780438115194Subjects--Topical Terms:

573171
Computer science.
Index Terms--Genre/Form:

554714
Electronic books.

Bounty Hunting : = A Dynamic Multiagent Task Allocation Mechanism.
LDR:03339ntm a2200361Ki 4500 001 916907
005 20180928111503.5
006 m o u
007 cr mn||||a|a||
008 190606s2018 xx obm 000 0 eng d
020 $a 9780438115194
035 $a (MiAaPQ)AAI10813507
035 $a (MiAaPQ)gmu:11739
035 $a AAI10813507
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Wicke, Drew. $3 1190775
245 1 0 $a Bounty Hunting : $b A Dynamic Multiagent Task Allocation Mechanism.
264 0 $c 2018
300 $a 1 online resource (135 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertation Abstracts International, Volume: 79-11(E), Section: B.
500 $a Adviser: Sean Luke.
502 $a Thesis (Ph.D.)--George Mason University, 2018.
504 $a Includes bibliographical references
520 $a Fundamental to many applications of multiagent systems is the problem of dynamically allocating tasks to agents that can solve them. Such dynamic task-allocation problems occur in a variety of real-life scenarios, ranging from automated delivery services to disaster response. The most popular current approach to solving this problem is based on auctioning tasks to the agents. This relies on nonintuitive assumptions such as an infinite money supply, exclusive allocation of tasks, and lack of incentive to complete the task. Such assumptions fail to hold in many real-world scenarios.
520 $a To address these issues, we propose an alternate system for multiagent task allocation inspired by the behavior of bounty hunters and bail bondsmen. In this model the bondsman assigns each task a reward called a bounty, and the bounty hunters greedily compete to finish the tasks. Upon completion, the bounty is awarded to the bounty hunter that completes the task first. The competition caused by nonexclusive allocation causes multiple agents to work on the same task, leading to an inefficient system. To solve this problem we study methods where the agents learn to split up the tasks so as to improve overall efficiency.
520 $a We study this technique and compare it to a number of exclusive methods including an auction-based approach. We first consider an environment where the agents must commit to tasks and find that bounty hunting is robust in a number of dynamic environments. We then relax task commitment to allow agents to change which task they are working on in order to address the problem of emergent and high-priority tasks. This new approach outperforms not only commitment-based bounty hunting methods, but also exclusive methods. Finally, we use bounty hunting as a heuristic to solve the dynamic traveling repairman problem, a dynamic task-allocation problem. We find that the bounty hunting approach Pareto- dominates the state-of-the-art heuristic, Nearest Neighbor. Finally, we give a demonstration of bounty hunting in cloud robotics and argue for its use in human-agent task allocation.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2018
538 $a Mode of access: World Wide Web
650 4 $a Computer science. $3 573171
650 4 $a Artificial intelligence. $3 559380
655 7 $a Electronic books. $2 local $3 554714
690 $a 0984
690 $a 0800
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a George Mason University. $b Computer Science. $3 1188721
773 0 $t Dissertation Abstracts International $g 79-11B(E).
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10813507 $z click for full text (PQDT)