No Cover Image

E-Thesis 369 views 143 downloads

Efficient Algorithms for Artificial Neural Networks and Explainable AI / HASSAN ESHKIKI

Swansea University Author: HASSAN ESHKIKI

  • 2023_Eshkiki_HG.final.63943.pdf

    PDF | E-Thesis – open access

    Copyright: The Author, Hassan G. Eshkiki, 2023.

    Download (8.21MB)

DOI (Published version): 10.23889/SUthesis.63943

Abstract

Artificial neural networks have allowed some remarkable progress in fields such as pattern recognition and computer vision. However, the increasing complexity of artificial neural networks presents a challenge for efficient computation. In this thesis, we first introduce a novel matrix multiplicatio...

Full description

Published: Swansea, Wales, UK 2023
Institution: Swansea University
Degree level: Doctoral
Degree name: Ph.D
Supervisor: Mora, Benjamin.
URI: https://cronfa.swan.ac.uk/Record/cronfa63943
Tags: Add Tag
No Tags, Be the first to tag this record!
first_indexed 2023-07-25T10:51:16Z
last_indexed 2023-07-25T10:51:16Z
id cronfa63943
recordtype RisThesis
fullrecord <?xml version="1.0" encoding="utf-8"?><rfc1807 xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema"><bib-version>v2</bib-version><id>63943</id><entry>2023-07-25</entry><title>Efficient Algorithms for Artificial Neural Networks and Explainable AI</title><swanseaauthors><author><sid>8e820926ba2c3f591969f1a5c9691ec0</sid><firstname>HASSAN</firstname><surname>ESHKIKI</surname><name>HASSAN ESHKIKI</name><active>true</active><ethesisStudent>false</ethesisStudent></author></swanseaauthors><date>2023-07-25</date><abstract>Artificial neural networks have allowed some remarkable progress in fields such as pattern recognition and computer vision. However, the increasing complexity of artificial neural networks presents a challenge for efficient computation. In this thesis, we first introduce a novel matrix multiplication method to reduce the complexity of artificial neural networks, where we demonstrate its suitability to compress fully connected layers of artificial neural networks. Our method outperforms other state-of-the-art methods when tested on standard publicly available datasets. This thesis then focuses on Explainable AI, which can be critical in fields like finance and medicine, as it can provide explanations for some decisions taken by sub-symbolic AI models behaving like a black box such as Artificial neural networks and transformation based learning approaches. We have also developed a new framework that facilitates the use of Explainable AI with tabular datasets. Our new framework Exmed, enables nonexpert users to prepare data, train models, and apply Explainable AI techniques effectively.Additionally, we propose a new algorithm that identifies the overall influence of input features and minimises the perturbations that alter the decision taken by a given model. Overall, this thesis introduces innovative and comprehensive techniques to enhance the efficiency of fully connected layers in artificial neural networks and provide a new approach to explain their decisions. These methods have significant practical applications in various fields, including portable medical devices.</abstract><type>E-Thesis</type><journal/><volume/><journalNumber/><paginationStart/><paginationEnd/><publisher/><placeOfPublication>Swansea, Wales, UK</placeOfPublication><isbnPrint/><isbnElectronic/><issnPrint/><issnElectronic/><keywords>ANN, DNN, ExAI, Matrix multiplication, Compressing, Counterfactual model</keywords><publishedDay>28</publishedDay><publishedMonth>6</publishedMonth><publishedYear>2023</publishedYear><publishedDate>2023-06-28</publishedDate><doi>10.23889/SUthesis.63943</doi><url/><notes/><college>COLLEGE NANME</college><CollegeCode>COLLEGE CODE</CollegeCode><institution>Swansea University</institution><supervisor>Mora, Benjamin.</supervisor><degreelevel>Doctoral</degreelevel><degreename>Ph.D</degreename><degreesponsorsfunders>EPSRC doctoral training grant</degreesponsorsfunders><apcterm/><funders/><projectreference/><lastEdited>2023-10-05T14:59:06.5692421</lastEdited><Created>2023-07-25T11:47:33.9805487</Created><path><level id="1">Faculty of Science and Engineering</level><level id="2">School of Mathematics and Computer Science - Computer Science</level></path><authors><author><firstname>HASSAN</firstname><surname>ESHKIKI</surname><order>1</order></author></authors><documents><document><filename>63943__28167__c1a364827cfc4daaac258bc71b71c7fc.pdf</filename><originalFilename>2023_Eshkiki_HG.final.63943.pdf</originalFilename><uploaded>2023-07-25T11:52:28.9058984</uploaded><type>Output</type><contentLength>8606913</contentLength><contentType>application/pdf</contentType><version>E-Thesis – open access</version><cronfaStatus>true</cronfaStatus><documentNotes>Copyright: The Author, Hassan G. Eshkiki, 2023.</documentNotes><copyrightCorrect>true</copyrightCorrect><language>eng</language></document></documents><OutputDurs/></rfc1807>
spelling v2 63943 2023-07-25 Efficient Algorithms for Artificial Neural Networks and Explainable AI 8e820926ba2c3f591969f1a5c9691ec0 HASSAN ESHKIKI HASSAN ESHKIKI true false 2023-07-25 Artificial neural networks have allowed some remarkable progress in fields such as pattern recognition and computer vision. However, the increasing complexity of artificial neural networks presents a challenge for efficient computation. In this thesis, we first introduce a novel matrix multiplication method to reduce the complexity of artificial neural networks, where we demonstrate its suitability to compress fully connected layers of artificial neural networks. Our method outperforms other state-of-the-art methods when tested on standard publicly available datasets. This thesis then focuses on Explainable AI, which can be critical in fields like finance and medicine, as it can provide explanations for some decisions taken by sub-symbolic AI models behaving like a black box such as Artificial neural networks and transformation based learning approaches. We have also developed a new framework that facilitates the use of Explainable AI with tabular datasets. Our new framework Exmed, enables nonexpert users to prepare data, train models, and apply Explainable AI techniques effectively.Additionally, we propose a new algorithm that identifies the overall influence of input features and minimises the perturbations that alter the decision taken by a given model. Overall, this thesis introduces innovative and comprehensive techniques to enhance the efficiency of fully connected layers in artificial neural networks and provide a new approach to explain their decisions. These methods have significant practical applications in various fields, including portable medical devices. E-Thesis Swansea, Wales, UK ANN, DNN, ExAI, Matrix multiplication, Compressing, Counterfactual model 28 6 2023 2023-06-28 10.23889/SUthesis.63943 COLLEGE NANME COLLEGE CODE Swansea University Mora, Benjamin. Doctoral Ph.D EPSRC doctoral training grant 2023-10-05T14:59:06.5692421 2023-07-25T11:47:33.9805487 Faculty of Science and Engineering School of Mathematics and Computer Science - Computer Science HASSAN ESHKIKI 1 63943__28167__c1a364827cfc4daaac258bc71b71c7fc.pdf 2023_Eshkiki_HG.final.63943.pdf 2023-07-25T11:52:28.9058984 Output 8606913 application/pdf E-Thesis – open access true Copyright: The Author, Hassan G. Eshkiki, 2023. true eng
title Efficient Algorithms for Artificial Neural Networks and Explainable AI
spellingShingle Efficient Algorithms for Artificial Neural Networks and Explainable AI
HASSAN ESHKIKI
title_short Efficient Algorithms for Artificial Neural Networks and Explainable AI
title_full Efficient Algorithms for Artificial Neural Networks and Explainable AI
title_fullStr Efficient Algorithms for Artificial Neural Networks and Explainable AI
title_full_unstemmed Efficient Algorithms for Artificial Neural Networks and Explainable AI
title_sort Efficient Algorithms for Artificial Neural Networks and Explainable AI
author_id_str_mv 8e820926ba2c3f591969f1a5c9691ec0
author_id_fullname_str_mv 8e820926ba2c3f591969f1a5c9691ec0_***_HASSAN ESHKIKI
author HASSAN ESHKIKI
author2 HASSAN ESHKIKI
format E-Thesis
publishDate 2023
institution Swansea University
doi_str_mv 10.23889/SUthesis.63943
college_str Faculty of Science and Engineering
hierarchytype
hierarchy_top_id facultyofscienceandengineering
hierarchy_top_title Faculty of Science and Engineering
hierarchy_parent_id facultyofscienceandengineering
hierarchy_parent_title Faculty of Science and Engineering
department_str School of Mathematics and Computer Science - Computer Science{{{_:::_}}}Faculty of Science and Engineering{{{_:::_}}}School of Mathematics and Computer Science - Computer Science
document_store_str 1
active_str 0
description Artificial neural networks have allowed some remarkable progress in fields such as pattern recognition and computer vision. However, the increasing complexity of artificial neural networks presents a challenge for efficient computation. In this thesis, we first introduce a novel matrix multiplication method to reduce the complexity of artificial neural networks, where we demonstrate its suitability to compress fully connected layers of artificial neural networks. Our method outperforms other state-of-the-art methods when tested on standard publicly available datasets. This thesis then focuses on Explainable AI, which can be critical in fields like finance and medicine, as it can provide explanations for some decisions taken by sub-symbolic AI models behaving like a black box such as Artificial neural networks and transformation based learning approaches. We have also developed a new framework that facilitates the use of Explainable AI with tabular datasets. Our new framework Exmed, enables nonexpert users to prepare data, train models, and apply Explainable AI techniques effectively.Additionally, we propose a new algorithm that identifies the overall influence of input features and minimises the perturbations that alter the decision taken by a given model. Overall, this thesis introduces innovative and comprehensive techniques to enhance the efficiency of fully connected layers in artificial neural networks and provide a new approach to explain their decisions. These methods have significant practical applications in various fields, including portable medical devices.
published_date 2023-06-28T14:59:08Z
_version_ 1778924229114200064
score 11.013148

Similar Items