東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Linked to FindBook

Google Book

Amazon

博客來

Towards Socially Interactive Agents: Learning Generative Models of Social Interactions Via Crowdsourcing.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Towards Socially Interactive Agents: Learning Generative Models of Social Interactions Via Crowdsourcing./
Author:	Feng, Dan.
Published:	Ann Arbor : ProQuest Dissertations & Theses, : 2020,
Description:	155 p.
Notes:	Source: Dissertations Abstracts International, Volume: 82-03, Section: B.
Contained By:	Dissertations Abstracts International82-03B.
Subject:	Computer science. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28088565
ISBN:	9798664788280

Towards Socially Interactive Agents: Learning Generative Models of Social Interactions Via Crowdsourcing.
Feng, Dan.

Towards Socially Interactive Agents: Learning Generative Models of Social Interactions Via Crowdsourcing. - Ann Arbor : ProQuest Dissertations & Theses, 2020 - 155 p.

Source: Dissertations Abstracts International, Volume: 82-03, Section: B.

Thesis (Ph.D.)--Northeastern University, 2020.

This item must not be sold to any third party vendors.

Modeling socially interactive agents (SIAs) has become critical in an increasingly wide range of applications, from large-scale social simulations to the design of embodied virtual humans. SIAs can be used to simulate the dynamics of human social behavior and engage with human users. The work introduced in this dissertation originates from building SIAs for pedagogical systems that support experiential learning. These systems offer various types of virtual worlds with simulated social environments that allow learners to interact safely with virtual characters in the context of an interactive story, explore the consequences of different actions, and then reflect on the experience. While these rich experience-based training systems show great promise, several challenges must be addressed. A key challenge is the extensive content requirement to build SIAs that support a rich, emergent, interactive experience with a wide range of action choices for the learner and the consequences they might bring. Traditionally, this content is hand-authored, which is insufficient and ineffective to sustain a flexible, varied learning experience. Ideally, a data-driven approach could alleviate the requirement of handcrafted content. However, because the focus of these SIA models is specific to a social scenario, there is often little to no data at scale available to train a model that could generate rich and situated human interactions. This raises the following questions: a) how to obtain data that reflects the richness of human social behaviors given a particular social scenario; b) what are the efficient ways of extracting domain knowledge from a limited amount of data; and c) how to build and evaluate generative models from the extracted knowledge. This dissertation attempts to tackle the aforementioned challenges of designing SIAs by exploring a hybrid data-driven approach that leverages both crowdsourcing and machine learning techniques. Specifically, this dissertation focuses on acquiring and modelling two key components of social knowledge-the action space and how the actions in the space connect to each other (e.g., interaction) using crowdsourcing. The results of this work demonstrate that crowdsourcing can be applied to collect a wide variety of actions and diverse knowledge that reflect the richness of social interactions in real life. Furthermore, this dissertation also sheds light on integrating crowdsourcing in a hybrid machine learning system to critique, assess, and refine the model.

ISBN: 9798664788280Subjects--Topical Terms:

523869
Computer science.
Subjects--Index Terms:

Crowdsourcing

Towards Socially Interactive Agents: Learning Generative Models of Social Interactions Via Crowdsourcing.
LDR:03664nmm a2200349 4500 001 2346343
005 20230315102144.5
006 m o d
007 cr#unu||||||||
008 241004s2020 ||||||||||||||||| ||eng d
020 $a 9798664788280
035 $a (MiAaPQ)AAI28088565
035 $a AAI28088565
040 $a MiAaPQ $c MiAaPQ
100 1 $a Feng, Dan. $3 3597877
245 1 0 $a Towards Socially Interactive Agents: Learning Generative Models of Social Interactions Via Crowdsourcing.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2020
300 $a 155 p.
500 $a Source: Dissertations Abstracts International, Volume: 82-03, Section: B.
500 $a Advisor: Marsella, Stacy.
502 $a Thesis (Ph.D.)--Northeastern University, 2020.
506 $a This item must not be sold to any third party vendors.
520 $a Modeling socially interactive agents (SIAs) has become critical in an increasingly wide range of applications, from large-scale social simulations to the design of embodied virtual humans. SIAs can be used to simulate the dynamics of human social behavior and engage with human users. The work introduced in this dissertation originates from building SIAs for pedagogical systems that support experiential learning. These systems offer various types of virtual worlds with simulated social environments that allow learners to interact safely with virtual characters in the context of an interactive story, explore the consequences of different actions, and then reflect on the experience. While these rich experience-based training systems show great promise, several challenges must be addressed. A key challenge is the extensive content requirement to build SIAs that support a rich, emergent, interactive experience with a wide range of action choices for the learner and the consequences they might bring. Traditionally, this content is hand-authored, which is insufficient and ineffective to sustain a flexible, varied learning experience. Ideally, a data-driven approach could alleviate the requirement of handcrafted content. However, because the focus of these SIA models is specific to a social scenario, there is often little to no data at scale available to train a model that could generate rich and situated human interactions. This raises the following questions: a) how to obtain data that reflects the richness of human social behaviors given a particular social scenario; b) what are the efficient ways of extracting domain knowledge from a limited amount of data; and c) how to build and evaluate generative models from the extracted knowledge. This dissertation attempts to tackle the aforementioned challenges of designing SIAs by exploring a hybrid data-driven approach that leverages both crowdsourcing and machine learning techniques. Specifically, this dissertation focuses on acquiring and modelling two key components of social knowledge-the action space and how the actions in the space connect to each other (e.g., interaction) using crowdsourcing. The results of this work demonstrate that crowdsourcing can be applied to collect a wide variety of actions and diverse knowledge that reflect the richness of social interactions in real life. Furthermore, this dissertation also sheds light on integrating crowdsourcing in a hybrid machine learning system to critique, assess, and refine the model.
590 $a School code: 0160.
650 4 $a Computer science. $3 523869
650 4 $a Artificial intelligence. $3 516317
653 $a Crowdsourcing
653 $a Socially Interactive Agents
690 $a 0984
690 $a 0800
710 2 $a Northeastern University. $b Computer Science. $3 1678818
773 0 $t Dissertations Abstracts International $g 82-03B.
790 $a 0160
791 $a Ph.D.
792 $a 2020
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28088565