Advanced search options

Advanced Search Options 🞨

Browse by author name (“Author name starts with…”).

Find ETDs with:

in
/  
in
/  
in
/  
in

Written in Published in Earliest date Latest date

Sorted by

Results per page:

Sorted by: relevance · author · university · dateNew search

Language: English

You searched for subject:(Chat bot). Showing records 1 – 2 of 2 total matches.

Search Limiters

Last 2 Years | English Only

No search limiters apply to these results.

▼ Search Limiters


KTH

1. Rinnarv, Jonathan. GANChat : A Generative Adversarial Network approach for chat bot learning.

Degree: Electrical Engineering and Computer Science (EECS), 2020, KTH

Recently a new method for training generative neural networks called Generative Adversarial Networks (GAN) has shown great results in the computer vision domain and shown potential in other generative machine learning tasks as well. GAN training is an adversarial training method where two neural networks compete and attempt to outperform each other, and in the process they both learn. In this thesis the effectiveness of GAN training is tested on conversational agents also called chat bots. To test this, current state-of-the-art training methods such as Maximum Likelihood Estimation (MLE) models are compared with GAN method trained models. Model performance was measured by closeness of the model distribution from the target distribution after training. This thesis shows that the GAN method performs worse the MLE in some scenarios but can outperform MLE in some cases.

Nyligen har en ny metod för att träna generativa neurala nätverk kallad Generative Adversarial Networks (GAN) visat bra resultat inom datorseendedomänen och visat potential inom andra maskininlärningsområden också GAN-träning är en träningsmetod där två neurala nätverk tävlar och försöker överträffa varandra, och i processen lär sig båda. I detta examensarbete har effektiviteten av GAN-träning testats på konversationsagenter, som också kallas Chat bots. För att testa det här jämfördes modeller tränade med nuvarande state-of- the-art träningsmetoder, så som Maximum likelihood-metoden (ML), med GAN-tränade modeller. Modellernas prestation mättes genom distans från modelldistribution till måldistribution efter träning. Det här examensarbetet visar att GAN-metoden presterar sämre än ML-metoden i vissa scenarier men kan överträffa ML i vissa fall.

Subjects/Keywords: Computer science Machine learning GAN chat bot; Computer Sciences; Datavetenskap (datalogi)

Record DetailsSimilar RecordsGoogle PlusoneFacebookTwitterCiteULikeMendeleyreddit

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6th Edition):

Rinnarv, J. (2020). GANChat : A Generative Adversarial Network approach for chat bot learning. (Thesis). KTH. Retrieved from http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-278143

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Chicago Manual of Style (16th Edition):

Rinnarv, Jonathan. “GANChat : A Generative Adversarial Network approach for chat bot learning.” 2020. Thesis, KTH. Accessed October 30, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-278143.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

MLA Handbook (7th Edition):

Rinnarv, Jonathan. “GANChat : A Generative Adversarial Network approach for chat bot learning.” 2020. Web. 30 Oct 2020.

Vancouver:

Rinnarv J. GANChat : A Generative Adversarial Network approach for chat bot learning. [Internet] [Thesis]. KTH; 2020. [cited 2020 Oct 30]. Available from: http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-278143.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Council of Science Editors:

Rinnarv J. GANChat : A Generative Adversarial Network approach for chat bot learning. [Thesis]. KTH; 2020. Available from: http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-278143

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation


University of Arizona

2. Schuetzler, Ryan M. Dynamic Interviewing Agents: Effects on Deception, Nonverbal Behavior, and Social Desirability .

Degree: 2015, University of Arizona

Virtual humans and other virtual agents are becoming more common in our everyday lives. Whether in the form of phone-based personal assistants or automated customer service systems, these technologies have begun to touch more of our activities. This research aims to understand how this technology affects the way we interact with our computer systems. Using a chat bot, I studied the way a conversational computer system affects the way people interact with and perceive automated interviewing systems in two different contexts. Study 1 examines the impact of a conversational agent on behavior during deception. It found that a conversational agent can have a powerful impact on people's perception of the system, resulting in individuals viewing the system as much more engaging and human. The conversational agent further results in a suppression of deception-related cues consistent with a more human-like interaction. Study 2 focuses on the effect of a conversational agent on socially desirable responding. Results of this study indicate that a conversational agent increases social desirability when the topic of the interview is sensitive, but has no effect when the questions are non-sensitive. The results of these two studies indicate that a conversational agent can change the way people interact with a computer system in substantial and meaningful ways. These studies represent a step toward understanding how conversational agents can shape the way we view and interact with computers. Advisors/Committee Members: Nunamaker, Jay F. Jr (advisor), Nunamaker, Jay F. Jr (committeemember), Burgoon, Judee K. (committeemember), Valacich, Joseph S. (committeemember).

Subjects/Keywords: conversational agent; deception; impression management; social desirability; Management Information Systems; chat bot

Record DetailsSimilar RecordsGoogle PlusoneFacebookTwitterCiteULikeMendeleyreddit

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6th Edition):

Schuetzler, R. M. (2015). Dynamic Interviewing Agents: Effects on Deception, Nonverbal Behavior, and Social Desirability . (Doctoral Dissertation). University of Arizona. Retrieved from http://hdl.handle.net/10150/556441

Chicago Manual of Style (16th Edition):

Schuetzler, Ryan M. “Dynamic Interviewing Agents: Effects on Deception, Nonverbal Behavior, and Social Desirability .” 2015. Doctoral Dissertation, University of Arizona. Accessed October 30, 2020. http://hdl.handle.net/10150/556441.

MLA Handbook (7th Edition):

Schuetzler, Ryan M. “Dynamic Interviewing Agents: Effects on Deception, Nonverbal Behavior, and Social Desirability .” 2015. Web. 30 Oct 2020.

Vancouver:

Schuetzler RM. Dynamic Interviewing Agents: Effects on Deception, Nonverbal Behavior, and Social Desirability . [Internet] [Doctoral dissertation]. University of Arizona; 2015. [cited 2020 Oct 30]. Available from: http://hdl.handle.net/10150/556441.

Council of Science Editors:

Schuetzler RM. Dynamic Interviewing Agents: Effects on Deception, Nonverbal Behavior, and Social Desirability . [Doctoral Dissertation]. University of Arizona; 2015. Available from: http://hdl.handle.net/10150/556441

.