{"id":{"repo_id":"middlesex","oai_identifier":"oai:repository.mdx.ac.uk:27z10q"},"canonical_url":"https://search.dev.ndltd.org/etd/middlesex/oai:repository.mdx.ac.uk:27z10q","repository":{"repo_id":"middlesex","name":"Middlesex University","base_url":"https://repository.mdx.ac.uk/oai2"},"display":{"title":"Unlocking the power of CNN models: enhancing training procedure with evolutionary based search for class-specific data augmentation","abstract":"Data augmentation (DA) is a critical technique for improving the generalization capa-bilities of Convolutional Neural Networks (CNNs) in image classification tasks. This re-search introduces an automated data augmentation framework that leverages a genetic algorithm (GA) to optimize class-specific augmentation strategies, aiming to enhance CNN performance across diverse datasets and architectures. Our framework integrates a GA to search for optimal DA strategies and a CNN as a fitness function to evalu-ate the effectiveness of each strategy. By employing this fitness-driven, evolutionary approach, the framework iteratively refines augmentation strategies over generations, selecting individuals that yield improved model performance. The framework was rigor-ously evaluated on a diverse set of datasets spanning medical, agricultural, and general domains. Results demonstrated significant improvements in validation loss, with reduc-tions of up to 58.56% compared to a non-augmented baseline and 54.28% compared to a baseline augmented with traditional methods. Additionally, our approach was benchmarked against a similar state-of-the-art framework from the literature. In this comparison, our method achieved a relative improvement of 210%, underscoring its su-perior efficiency in optimizing DA strategies. These findings highlight the framework’s robustness and potential to serve as a universal solution for enhancing CNN classifi-cation performance. The framework provides a practical, adaptable tool that can be seamlessly integrated into existing workflows to improve model performance, particu-larly in domains with limited data and contributes to the growing field of automated data augmentation.","abstract_html":"Data augmentation (DA) is a critical technique for improving the generalization capa-bilities of Convolutional Neural Networks (CNNs) in image classification tasks. This re-search introduces an automated data augmentation framework that leverages a genetic algorithm (GA) to optimize class-specific augmentation strategies, aiming to enhance CNN performance across diverse datasets and architectures. Our framework integrates a GA to search for optimal DA strategies and a CNN as a fitness function to evalu-ate the effectiveness of each strategy. By employing this fitness-driven, evolutionary approach, the framework iteratively refines augmentation strategies over generations, selecting individuals that yield improved model performance. The framework was rigor-ously evaluated on a diverse set of datasets spanning medical, agricultural, and general domains. Results demonstrated significant improvements in validation loss, with reduc-tions of up to 58.56% compared to a non-augmented baseline and 54.28% compared to a baseline augmented with traditional methods. Additionally, our approach was benchmarked against a similar state-of-the-art framework from the literature. In this comparison, our method achieved a relative improvement of 210%, underscoring its su-perior efficiency in optimizing DA strategies. These findings highlight the framework’s robustness and potential to serve as a universal solution for enhancing CNN classifi-cation performance. The framework provides a practical, adaptable tool that can be seamlessly integrated into existing workflows to improve model performance, particu-larly in domains with limited data and contributes to the growing field of automated data augmentation.","abstract_has_math":false,"creators":["Marc, S.T."],"institution":"Middlesex University","degree_name":"PhD","degree_level":"PhD thesis","degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025","date_published":"2025","updated_at":"2026-07-24T03:03:20Z","subjects":[],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["oai:repository.mdx.ac.uk:27z10q"],"render_values":[{"text":"oai:repository.mdx.ac.uk:27z10q","href":null,"code":true}]}]},"links":{"outbound_url":null,"outbound_label":null,"outbound_source":null},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Marc, S.T."]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2025"]},{"key":"dc:date.issued","label":"Date","values":["2025"]},{"key":"dc:publisher","label":"Institution","values":["Middlesex University Research Repository"]},{"key":"dc:publisher.department","label":"Dc Publisher Department","values":["Computer Science","Science and Technology"]},{"key":"dc:publisher.institution","label":"Dc Publisher Institution","values":["Middlesex University"]},{"key":"dc:relation","label":"Dc Relation","values":["https://repository.mdx.ac.uk/item/27z10q"]},{"key":"dc:relation.isreferencedby","label":"Dc Relation Isreferencedby","values":["https://repository.mdx.ac.uk/item/27z10q"]},{"key":"dc:type","label":"Dc Type","values":["Thesis or dissertation"]},{"key":"dc:type.qualificationlevel","label":"Dc Type Qualificationlevel","values":["PhD thesis"]},{"key":"dc:type.qualificationname","label":"Dc Type Qualificationname","values":["PhD"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["oai:repository.mdx.ac.uk:27z10q"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://repository.mdx.ac.uk/download/742b6e1846936c41113941c01a90beae24c800a155a68975464850e340b445fe/5136932/STMarc%20thesis.pdf"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Data augmentation (DA) is a critical technique for improving the generalization capa-bilities of Convolutional Neural Networks (CNNs) in image classification tasks. This re-search introduces an automated data augmentation framework that leverages a genetic algorithm (GA) to optimize class-specific augmentation strategies, aiming to enhance CNN performance across diverse datasets and architectures. Our framework integrates a GA to search for optimal DA strategies and a CNN as a fitness function to evalu-ate the effectiveness of each strategy. By employing this fitness-driven, evolutionary approach, the framework iteratively refines augmentation strategies over generations, selecting individuals that yield improved model performance. The framework was rigor-ously evaluated on a diverse set of datasets spanning medical, agricultural, and general domains. Results demonstrated significant improvements in validation loss, with reduc-tions of up to 58.56% compared to a non-augmented baseline and 54.28% compared to a baseline augmented with traditional methods. Additionally, our approach was benchmarked against a similar state-of-the-art framework from the literature. In this comparison, our method achieved a relative improvement of 210%, underscoring its su-perior efficiency in optimizing DA strategies. These findings highlight the framework’s robustness and potential to serve as a universal solution for enhancing CNN classifi-cation performance. The framework provides a practical, adaptable tool that can be seamlessly integrated into existing workflows to improve model performance, particu-larly in domains with limited data and contributes to the growing field of automated data augmentation."]},{"key":"dc:description.abstract","label":"Abstract","values":["Data augmentation (DA) is a critical technique for improving the generalization capa-bilities of Convolutional Neural Networks (CNNs) in image classification tasks. This re-search introduces an automated data augmentation framework that leverages a genetic algorithm (GA) to optimize class-specific augmentation strategies, aiming to enhance CNN performance across diverse datasets and architectures. Our framework integrates a GA to search for optimal DA strategies and a CNN as a fitness function to evalu-ate the effectiveness of each strategy. By employing this fitness-driven, evolutionary approach, the framework iteratively refines augmentation strategies over generations, selecting individuals that yield improved model performance. The framework was rigor-ously evaluated on a diverse set of datasets spanning medical, agricultural, and general domains. Results demonstrated significant improvements in validation loss, with reduc-tions of up to 58.56% compared to a non-augmented baseline and 54.28% compared to a baseline augmented with traditional methods. Additionally, our approach was benchmarked against a similar state-of-the-art framework from the literature. In this comparison, our method achieved a relative improvement of 210%, underscoring its su-perior efficiency in optimizing DA strategies. These findings highlight the framework’s robustness and potential to serve as a universal solution for enhancing CNN classifi-cation performance. The framework provides a practical, adaptable tool that can be seamlessly integrated into existing workflows to improve model performance, particu-larly in domains with limited data and contributes to the growing field of automated data augmentation."]},{"key":"dc:title","label":"Title","values":["Unlocking the power of CNN models: enhancing training procedure with evolutionary based search for class-specific data augmentation"]}]}],"canonical_facts":{"dc:creator":["Marc, S.T."],"dc:date":["2025"],"dc:date.issued":["2025"],"dc:description":["Data augmentation (DA) is a critical technique for improving the generalization capa-bilities of Convolutional Neural Networks (CNNs) in image classification tasks. This re-search introduces an automated data augmentation framework that leverages a genetic algorithm (GA) to optimize class-specific augmentation strategies, aiming to enhance CNN performance across diverse datasets and architectures. Our framework integrates a GA to search for optimal DA strategies and a CNN as a fitness function to evalu-ate the effectiveness of each strategy. By employing this fitness-driven, evolutionary approach, the framework iteratively refines augmentation strategies over generations, selecting individuals that yield improved model performance. The framework was rigor-ously evaluated on a diverse set of datasets spanning medical, agricultural, and general domains. Results demonstrated significant improvements in validation loss, with reduc-tions of up to 58.56% compared to a non-augmented baseline and 54.28% compared to a baseline augmented with traditional methods. Additionally, our approach was benchmarked against a similar state-of-the-art framework from the literature. In this comparison, our method achieved a relative improvement of 210%, underscoring its su-perior efficiency in optimizing DA strategies. These findings highlight the framework’s robustness and potential to serve as a universal solution for enhancing CNN classifi-cation performance. The framework provides a practical, adaptable tool that can be seamlessly integrated into existing workflows to improve model performance, particu-larly in domains with limited data and contributes to the growing field of automated data augmentation."],"dc:description.abstract":["Data augmentation (DA) is a critical technique for improving the generalization capa-bilities of Convolutional Neural Networks (CNNs) in image classification tasks. This re-search introduces an automated data augmentation framework that leverages a genetic algorithm (GA) to optimize class-specific augmentation strategies, aiming to enhance CNN performance across diverse datasets and architectures. Our framework integrates a GA to search for optimal DA strategies and a CNN as a fitness function to evalu-ate the effectiveness of each strategy. By employing this fitness-driven, evolutionary approach, the framework iteratively refines augmentation strategies over generations, selecting individuals that yield improved model performance. The framework was rigor-ously evaluated on a diverse set of datasets spanning medical, agricultural, and general domains. Results demonstrated significant improvements in validation loss, with reduc-tions of up to 58.56% compared to a non-augmented baseline and 54.28% compared to a baseline augmented with traditional methods. Additionally, our approach was benchmarked against a similar state-of-the-art framework from the literature. In this comparison, our method achieved a relative improvement of 210%, underscoring its su-perior efficiency in optimizing DA strategies. These findings highlight the framework’s robustness and potential to serve as a universal solution for enhancing CNN classifi-cation performance. The framework provides a practical, adaptable tool that can be seamlessly integrated into existing workflows to improve model performance, particu-larly in domains with limited data and contributes to the growing field of automated data augmentation."],"dc:identifier":["oai:repository.mdx.ac.uk:27z10q"],"dc:identifier.uri":["https://repository.mdx.ac.uk/download/742b6e1846936c41113941c01a90beae24c800a155a68975464850e340b445fe/5136932/STMarc%20thesis.pdf"],"dc:publisher":["Middlesex University Research Repository"],"dc:publisher.department":["Computer Science","Science and Technology"],"dc:publisher.institution":["Middlesex University"],"dc:relation":["https://repository.mdx.ac.uk/item/27z10q"],"dc:relation.isreferencedby":["https://repository.mdx.ac.uk/item/27z10q"],"dc:title":["Unlocking the power of CNN models: enhancing training procedure with evolutionary based search for class-specific data augmentation"],"dc:type":["Thesis or dissertation"],"dc:type.qualificationlevel":["PhD thesis"],"dc:type.qualificationname":["PhD"]},"updated_at":"2026-07-24T03:03:20Z"}