{"id":{"repo_id":"mcmaster","oai_identifier":"oai:macsphere.mcmaster.ca:11375/33063"},"canonical_url":"https://search.ndltd.org/etd/mcmaster/oai:macsphere.mcmaster.ca:11375/33063","repository":{"repo_id":"mcmaster","name":"McMaster University","base_url":"https://macsphere.mcmaster.ca/server/oai/request"},"display":{"title":"DESIGN AND OPTIMIZATION OF ELECTROMAGNETIC CARPET CLOAK USING MACHINE LEARNING","abstract":"Electromagnetic carpet cloaks are designed to restore the reflection behavior of a flat conducting surface in the presence of geometric irregularities. While transformation-optics based approaches provide a rigorous theoretical foundation for such cloaks, the resulting refractive index distributions often have their values going beyond a realistic range, which makes the physical implementation of such carpet cloaks not feasible. This thesis presents a hybrid framework that integrates analytical modeling, numerical reconstruction, and machine learning based refractive index generation for two-dimensional electromagnetic carpet cloaking. An analytically defined refractive index profile is firstly investigated and shown to offer only limited reflection restoration when applied to a prescribed cloak geometry. A physics-based reference refractive index distribution is then reconstructed using conformal mapping techniques, capturing the essential spatial features of an ideal carpet cloak without requiring electromagnetic simulations. Using this reference distribution as training data, a supervised machine-learning regression model is developed to learn the spatial dependence of the refractive index within a fixed cloak geometry. Rather than directly reproducing the reference distribution, the learning process reformulates it into a smooth and bounded refractive-index profile while preserving the reflection characteristics of an ideal flat conducting surface. The resulting machine-learning-generated profile operates within a reduced refractive-index range of approximately 1.0-1.33, significantly lower than that required by the reference transformation-optics design. Finite-element simulations demonstrate that the machine-learning-based refractive-index profile achieves near-ideal flat-surface reflection behavior and significantly reduces the mean-squared error compared with both the analytical design and the original reference distribution. These results show that data-driven refractive-index generation, guided by physics-based reference models, can improve electromagnetic cloaking performance while reducing refractive-index requirements within fixed geometries.","abstract_html":"Electromagnetic carpet cloaks are designed to restore the reflection behavior of a flat conducting surface in the presence of geometric irregularities. While transformation-optics based approaches provide a rigorous theoretical foundation for such cloaks, the resulting refractive index distributions often have their values going beyond a realistic range, which makes the physical implementation of such carpet cloaks not feasible. This thesis presents a hybrid framework that integrates analytical modeling, numerical reconstruction, and machine learning based refractive index generation for two-dimensional electromagnetic carpet cloaking. An analytically defined refractive index profile is firstly investigated and shown to offer only limited reflection restoration when applied to a prescribed cloak geometry. A physics-based reference refractive index distribution is then reconstructed using conformal mapping techniques, capturing the essential spatial features of an ideal carpet cloak without requiring electromagnetic simulations. Using this reference distribution as training data, a supervised machine-learning regression model is developed to learn the spatial dependence of the refractive index within a fixed cloak geometry. Rather than directly reproducing the reference distribution, the learning process reformulates it into a smooth and bounded refractive-index profile while preserving the reflection characteristics of an ideal flat conducting surface. The resulting machine-learning-generated profile operates within a reduced refractive-index range of approximately 1.0-1.33, significantly lower than that required by the reference transformation-optics design. Finite-element simulations demonstrate that the machine-learning-based refractive-index profile achieves near-ideal flat-surface reflection behavior and significantly reduces the mean-squared error compared with both the analytical design and the original reference distribution. These results show that data-driven refractive-index generation, guided by physics-based reference models, can improve electromagnetic cloaking performance while reducing refractive-index requirements within fixed geometries.","abstract_has_math":false,"creators":["Kayser, Sayda Fahmida"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":"Electrical and Computer Engineering","school":null,"contributors":[],"advisors":["Li, Xun"],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T19:38:20Z","subjects":[],"languages":["en_US"],"rights":["Attribution-NonCommercial-NoDerivs 2.5 Canada"],"rights_urls":["http://creativecommons.org/licenses/by-nc-nd/2.5/ca/"],"identifier_entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://doi.org/10.71548/20747"],"render_values":[{"text":"https://doi.org/10.71548/20747","href":"https://doi.org/10.71548/20747","code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/11375/33063","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/11375/33063","record_label":"Repository record","record_kind":"handle","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://macsphere.mcmaster.ca/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Amacsphere.mcmaster.ca%3A11375%2F33063","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Li, Xun"]},{"key":"dc:contributor.department","label":"Department","values":["Electrical and Computer Engineering"]},{"key":"dc:creator","label":"Author","values":["Kayser, Sayda Fahmida"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-04-17T18:27:26Z"]},{"key":"dc:date.issued","label":"Date","values":["2026"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en_US"]},{"key":"dc:rights","label":"Dc Rights","values":["Attribution-NonCommercial-NoDerivs 2.5 Canada"]},{"key":"dc:rights.uri","label":"Rights URI","values":["http://creativecommons.org/licenses/by-nc-nd/2.5/ca/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/11375/33063","https://doi.org/10.71548/20747"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Electromagnetic carpet cloaks are designed to restore the reflection behavior of a flat conducting surface in the presence of geometric irregularities. While transformation-optics based approaches provide a rigorous theoretical foundation for such cloaks, the resulting refractive index distributions often have their values going beyond a realistic range, which makes the physical implementation of such carpet cloaks not feasible. This thesis presents a hybrid framework that integrates analytical modeling, numerical reconstruction, and machine learning based refractive index generation for two-dimensional electromagnetic carpet cloaking. An analytically defined refractive index profile is firstly investigated and shown to offer only limited reflection restoration when applied to a prescribed cloak geometry. A physics-based reference refractive index distribution is then reconstructed using conformal mapping techniques, capturing the essential spatial features of an ideal carpet cloak without requiring electromagnetic simulations. Using this reference distribution as training data, a supervised machine-learning regression model is developed to learn the spatial dependence of the refractive index within a fixed cloak geometry. Rather than directly reproducing the reference distribution, the learning process reformulates it into a smooth and bounded refractive-index profile while preserving the reflection characteristics of an ideal flat conducting surface. The resulting machine-learning-generated profile operates within a reduced refractive-index range of approximately 1.0-1.33, significantly lower than that required by the reference transformation-optics design. Finite-element simulations demonstrate that the machine-learning-based refractive-index profile achieves near-ideal flat-surface reflection behavior and significantly reduces the mean-squared error compared with both the analytical design and the original reference distribution. These results show that data-driven refractive-index generation, guided by physics-based reference models, can improve electromagnetic cloaking performance while reducing refractive-index requirements within fixed geometries."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Master of Applied Science (MASc)"]},{"key":"dc:title","label":"Title","values":["DESIGN AND OPTIMIZATION OF ELECTROMAGNETIC CARPET CLOAK USING MACHINE LEARNING"]},{"key":"dc:title.alternative","label":"Alternative Title","values":["DESIGN OPTIMIZATION OF CARPET CLOAK VIA MACHINE LEARNING"]},{"key":"others:access_status","label":"Others Access Status","values":["open.access"]}]}],"canonical_facts":{"dc:contributor.advisor":["Li, Xun"],"dc:contributor.department":["Electrical and Computer Engineering"],"dc:creator":["Kayser, Sayda Fahmida"],"dc:date.accessioned":["2026-04-17T18:27:26Z"],"dc:date.issued":["2026"],"dc:description.abstract":["Electromagnetic carpet cloaks are designed to restore the reflection behavior of a flat conducting surface in the presence of geometric irregularities. While transformation-optics based approaches provide a rigorous theoretical foundation for such cloaks, the resulting refractive index distributions often have their values going beyond a realistic range, which makes the physical implementation of such carpet cloaks not feasible. This thesis presents a hybrid framework that integrates analytical modeling, numerical reconstruction, and machine learning based refractive index generation for two-dimensional electromagnetic carpet cloaking. An analytically defined refractive index profile is firstly investigated and shown to offer only limited reflection restoration when applied to a prescribed cloak geometry. A physics-based reference refractive index distribution is then reconstructed using conformal mapping techniques, capturing the essential spatial features of an ideal carpet cloak without requiring electromagnetic simulations. Using this reference distribution as training data, a supervised machine-learning regression model is developed to learn the spatial dependence of the refractive index within a fixed cloak geometry. Rather than directly reproducing the reference distribution, the learning process reformulates it into a smooth and bounded refractive-index profile while preserving the reflection characteristics of an ideal flat conducting surface. The resulting machine-learning-generated profile operates within a reduced refractive-index range of approximately 1.0-1.33, significantly lower than that required by the reference transformation-optics design. Finite-element simulations demonstrate that the machine-learning-based refractive-index profile achieves near-ideal flat-surface reflection behavior and significantly reduces the mean-squared error compared with both the analytical design and the original reference distribution. These results show that data-driven refractive-index generation, guided by physics-based reference models, can improve electromagnetic cloaking performance while reducing refractive-index requirements within fixed geometries."],"dc:description.degree":["Master of Applied Science (MASc)"],"dc:identifier.uri":["https://hdl.handle.net/11375/33063","https://doi.org/10.71548/20747"],"dc:language.iso":["en_US"],"dc:rights":["Attribution-NonCommercial-NoDerivs 2.5 Canada"],"dc:rights.uri":["http://creativecommons.org/licenses/by-nc-nd/2.5/ca/"],"dc:title":["DESIGN AND OPTIMIZATION OF ELECTROMAGNETIC CARPET CLOAK USING MACHINE LEARNING"],"dc:title.alternative":["DESIGN OPTIMIZATION OF CARPET CLOAK VIA MACHINE LEARNING"],"dc:type":["Thesis"],"others:access_status":["open.access"]},"updated_at":"2026-11-18T19:38:20Z"}