國立虎尾科技大學 |

Augmenting Artificial Intelligence Design and Performance With Human Physiological and Behavioral Information /

Record Type:	Language materials, printed : Monograph/item
Title/Author:	Augmenting Artificial Intelligence Design and Performance With Human Physiological and Behavioral Information // Joseph Philip Distefano.
Author:	Distefano, Joseph Philip,
Description:	1 electronic resource (133 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 86-03, Section: B.
Contained By:	Dissertations Abstracts International86-03B.
Subject:	Mechanical engineering. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31557342
ISBN:	9798384091455

Augmenting Artificial Intelligence Design and Performance With Human Physiological and Behavioral Information /
Distefano, Joseph Philip,

Augmenting Artificial Intelligence Design and Performance With Human Physiological and Behavioral Information /Joseph Philip Distefano. - 1 electronic resource (133 pages)

Source: Dissertations Abstracts International, Volume: 86-03, Section: B.

Recent advances in human artificial intelligence collaboration, complex multirobot systems, and human-in-the-loop learning have enabled the development and application of new systems in a plethora of different fields including healthcare, military operations, autonomous transportation, and manufacturing. In these systems, it is essential that both AI and human factors are considered to ensure mutual benefit and shared autonomy. In this direction, both an understanding of how AI can be designed to benefit humans, and, how humans can benefit the design of AI must be explored.It is required to have an understanding of human cognition and behavior to design an effective AI that can benefit a human's performance. For example, if an AI does not provide enough input when the human workload is high, the human performance will be lower and trust in the AI will decrease. Furthermore, through the study of human cognition and behavior, the design of an adaptive personalized AI can increase the performance of the system.On the other hand, an AI system lacks the visual and information processing and prioritization that humans possess. Human attention provides a wealth of information about the environment and actions that need to be taken to complete a mission. Through the utilization of predicted human attention in training AI, we can augment the training and performance.In this regard, this thesis explores two avenues: 1) Measuring and quantifying human cognition and behavior during human-AI collaboration in multi-robot systems. 2) Utilization of predicted human attention to augment AI algorithms. Human swarm interaction is utilized to study human cognition and behavior during human-AI collaboration. We run multiple human subject studies where a human is placed in a supervisory role and must control multiple platoons with two variations: The first being varying difficulty in the environment and the second being different levels of feedback and compliance. Through this study, we use physiological measurements to estimate and predict human intent, task difficulty, workload, individual differences, trust, and cognition during non-compliance. Next, predicted attention from humans is used to augment human-in-the-loop reinforcement learning and imitation learning in the Atari environment.

English

ISBN: 9798384091455Subjects--Topical Terms:

557493
Mechanical engineering.
Subjects--Index Terms:

Human cognition

Augmenting Artificial Intelligence Design and Performance With Human Physiological and Behavioral Information /
LDR:03969nam a22004693i 4500 001 1157864
005 20250603111431.5
006 m o d
007 cr|nu||||||||
008 250804s2024 miu||||||m |||||||eng d
020 $a 9798384091455
035 $a (MiAaPQD)AAI31557342
035 $a AAI31557342
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Distefano, Joseph Philip, $e author. $3 1484147
245 1 0 $a Augmenting Artificial Intelligence Design and Performance With Human Physiological and Behavioral Information / $c Joseph Philip Distefano.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2024
300 $a 1 electronic resource (133 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: Dissertations Abstracts International, Volume: 86-03, Section: B.
500 $a Advisors: Esfahani, Ehsan T. Committee members: Chowdhury, Souma; St. Pierre, Ryan; Dantu, Karthik.
502 $b Ph.D. $c State University of New York at Buffalo $d 2024.
520 $a Recent advances in human artificial intelligence collaboration, complex multirobot systems, and human-in-the-loop learning have enabled the development and application of new systems in a plethora of different fields including healthcare, military operations, autonomous transportation, and manufacturing. In these systems, it is essential that both AI and human factors are considered to ensure mutual benefit and shared autonomy. In this direction, both an understanding of how AI can be designed to benefit humans, and, how humans can benefit the design of AI must be explored.It is required to have an understanding of human cognition and behavior to design an effective AI that can benefit a human's performance. For example, if an AI does not provide enough input when the human workload is high, the human performance will be lower and trust in the AI will decrease. Furthermore, through the study of human cognition and behavior, the design of an adaptive personalized AI can increase the performance of the system.On the other hand, an AI system lacks the visual and information processing and prioritization that humans possess. Human attention provides a wealth of information about the environment and actions that need to be taken to complete a mission. Through the utilization of predicted human attention in training AI, we can augment the training and performance.In this regard, this thesis explores two avenues: 1) Measuring and quantifying human cognition and behavior during human-AI collaboration in multi-robot systems. 2) Utilization of predicted human attention to augment AI algorithms. Human swarm interaction is utilized to study human cognition and behavior during human-AI collaboration. We run multiple human subject studies where a human is placed in a supervisory role and must control multiple platoons with two variations: The first being varying difficulty in the environment and the second being different levels of feedback and compliance. Through this study, we use physiological measurements to estimate and predict human intent, task difficulty, workload, individual differences, trust, and cognition during non-compliance. Next, predicted attention from humans is used to augment human-in-the-loop reinforcement learning and imitation learning in the Atari environment.
546 $a English
590 $a School code: 0656
650 4 $a Mechanical engineering. $3 557493
650 4 $a Neurosciences. $3 593561
650 4 $a Computer engineering. $3 569006
650 4 $a Robotics. $3 561941
653 $a Human cognition
653 $a Human-AI collaboration
653 $a Human-in-the-loop systems
653 $a Human-swarm interaction
653 $a Physiological information
653 $a Reinforcement learning
690 $a 0548
690 $a 0800
690 $a 0317
690 $a 0464
690 $a 0771
710 2 $a State University of New York at Buffalo. $b Mechanical and Aerospace Engineering. $3 845600
720 1 $a Esfahani, Ehsan T. $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 86-03B.
790 $a 0656
791 $a Ph.D.
792 $a 2024
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31557342