{"id":{"repo_id":"duquesne","oai_identifier":"oai:dsc.duq.edu:etd-3147"},"canonical_url":"https://search.ndltd.org/etd/duquesne/oai:dsc.duq.edu:etd-3147","repository":{"repo_id":"duquesne","name":"Duquesne","base_url":"https://dsc.duq.edu/do/oai/"},"display":{"title":"Mathematical Models Yield Insights into CNNs: Applications in Natural Image Restoration and Population Genetics","abstract":"Due to a rise in computational power, machine learning (ML) methods have become the state-of-the-art in a variety of fields. Known to be black-box approaches, however, these methods are oftentimes not well understood. In this work, we utilize our understanding of model-based approaches to derive insights into Convolutional Neural Networks (CNNs). In the field of Natural Image Restoration, we focus on the image denoising problem. Recent work have demonstrated the potential of mathematically motivated CNN architectures that learn both `geometric' and nonlinear higher order features and corresponding regularizers. We extend this work by showing that not only can geometric features and corresponding regularizers be learned through ML but also be used to boost the performance of other CNN approaches. In the field of population genetics, we focus on the problem of detecting selective sweeps within genomes where various CNN based approaches have been proposed. We give an overview of these approaches and show how their method of prediction is based on a measure quite similar to the Garud's H test statistic.","abstract_html":"Due to a rise in computational power, machine learning (ML) methods have become the state-of-the-art in a variety of fields. Known to be black-box approaches, however, these methods are oftentimes not well understood. In this work, we utilize our understanding of model-based approaches to derive insights into Convolutional Neural Networks (CNNs). In the field of Natural Image Restoration, we focus on the image denoising problem. Recent work have demonstrated the potential of mathematically motivated CNN architectures that learn both `geometric&#x27; and nonlinear higher order features and corresponding regularizers. We extend this work by showing that not only can geometric features and corresponding regularizers be learned through ML but also be used to boost the performance of other CNN approaches. In the field of population genetics, we focus on the problem of detecting selective sweeps within genomes where various CNN based approaches have been proposed. We give an overview of these approaches and show how their method of prediction is based on a measure quite similar to the Garud&#x27;s H test statistic.","abstract_has_math":false,"creators":["Cecil, Ryan"],"institution":null,"degree_name":"MS","degree_level":"One-year Embargo","degree_discipline":"Computational Mathematics","degree_department":null,"school":null,"contributors":["Lauren Sugden","Stacey Levine"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":null,"date_issued":"","date_published":null,"updated_at":"2026-11-19T07:51:57Z","subjects":["Image Processing","Statistical Genetics","Machine Learning","Artificial Intelligence and Robotics","Biostatistics","Data Science","Numerical Analysis and Scientific Computing","Partial Differential Equations"],"languages":["English"],"rights":[],"rights_urls":[],"identifier_entries":[],"person_identifier_entries":[]},"links":{"outbound_url":"https://dsc.duq.edu/etd/2177","outbound_label":"Repository record","outbound_source":"dc:identifier","record_url":"https://dsc.duq.edu/etd/2177","record_label":"Repository record","record_kind":"record","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://dsc.duq.edu/do/oai/?verb=GetRecord&metadataPrefix=oai_etdms&identifier=oai%3Adsc.duq.edu%3Aetd-3147","prefix":"oai_etdms"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Lauren Sugden","Stacey Levine"]},{"key":"dc:creator","label":"Author","values":["Cecil, Ryan"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.available","label":"Dc Date Available","values":["2023-09-21T07:00:00Z"]},{"key":"dc:date.created","label":"Dc Date Created","values":["2022-08-13T07:00:00Z"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computational Mathematics"]},{"key":"thesis:degree_level","label":"Degree Level","values":["One-year Embargo"]},{"key":"thesis:degree_name","label":"Degree Name","values":["MS"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Image Processing","Statistical Genetics","Machine Learning","Artificial Intelligence and Robotics","Biostatistics","Data Science","Numerical Analysis and Scientific Computing","Partial Differential Equations"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["English"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://dsc.duq.edu/etd/2177"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Due to a rise in computational power, machine learning (ML) methods have become the state-of-the-art in a variety of fields. Known to be black-box approaches, however, these methods are oftentimes not well understood. In this work, we utilize our understanding of model-based approaches to derive insights into Convolutional Neural Networks (CNNs). In the field of Natural Image Restoration, we focus on the image denoising problem. Recent work have demonstrated the potential of mathematically motivated CNN architectures that learn both `geometric' and nonlinear higher order features and corresponding regularizers. We extend this work by showing that not only can geometric features and corresponding regularizers be learned through ML but also be used to boost the performance of other CNN approaches. In the field of population genetics, we focus on the problem of detecting selective sweeps within genomes where various CNN based approaches have been proposed. We give an overview of these approaches and show how their method of prediction is based on a measure quite similar to the Garud's H test statistic."]},{"key":"dc:title","label":"Title","values":["Mathematical Models Yield Insights into CNNs: Applications in Natural Image Restoration and Population Genetics"]}]}],"canonical_facts":{"dc:contributor":["Lauren Sugden","Stacey Levine"],"dc:creator":["Cecil, Ryan"],"dc:date.available":["2023-09-21T07:00:00Z"],"dc:date.created":["2022-08-13T07:00:00Z"],"dc:description.abstract":["Due to a rise in computational power, machine learning (ML) methods have become the state-of-the-art in a variety of fields. Known to be black-box approaches, however, these methods are oftentimes not well understood. In this work, we utilize our understanding of model-based approaches to derive insights into Convolutional Neural Networks (CNNs). In the field of Natural Image Restoration, we focus on the image denoising problem. Recent work have demonstrated the potential of mathematically motivated CNN architectures that learn both `geometric' and nonlinear higher order features and corresponding regularizers. We extend this work by showing that not only can geometric features and corresponding regularizers be learned through ML but also be used to boost the performance of other CNN approaches. In the field of population genetics, we focus on the problem of detecting selective sweeps within genomes where various CNN based approaches have been proposed. We give an overview of these approaches and show how their method of prediction is based on a measure quite similar to the Garud's H test statistic."],"dc:identifier":["https://dsc.duq.edu/etd/2177"],"dc:language":["English"],"dc:subject":["Image Processing","Statistical Genetics","Machine Learning","Artificial Intelligence and Robotics","Biostatistics","Data Science","Numerical Analysis and Scientific Computing","Partial Differential Equations"],"dc:title":["Mathematical Models Yield Insights into CNNs: Applications in Natural Image Restoration and Population Genetics"],"thesis:degree_discipline":["Computational Mathematics"],"thesis:degree_level":["One-year Embargo"],"thesis:degree_name":["MS"]},"updated_at":"2026-11-19T07:51:57Z"}