{"id":{"repo_id":"msu","oai_identifier":"oai:d.lib.msu.edu:etd_19404"},"canonical_url":"https://search.ndltd.org/etd/msu/oai:d.lib.msu.edu:etd_19404","repository":{"repo_id":"msu","name":"Michigan State University","base_url":"https://d.lib.msu.edu/oai"},"display":{"title":"Representation learning and image synthesis for deep face recognition","abstract":"\"Face recognition has been advanced a lot in recent years thanks to the development of deep neural networks. The large intra-class variations in pose, illumination, and expression are the long-standing challenges. Learning a discriminative representation that is robust to these variations is the key. In the scenarios of profile pose or long-tail training data, image or feature-level data augmentation is needed. This dissertation presents three different methods to solve these problems. First, we explore a multi-task Convolutional Neural Network (CNN) that aims to leverage side tasks to improve representation learning. A pose-directed multi-task CNN is introduced to better handle pose variation. The proposed framework is effective in pose-invariant face recognition. Second, we propose a Face Frontalization-Generative Adversarial Network (FF-GAN) that can generate a frontal face even from an input image with extreme profile pose. FF-GAN handles pose variation from the perspective of image-level data augmentation. Multiple loss functions are proposed to achieve large-pose face frontalization. The proposed approach is evaluated on various tasks including face reconstruction, landmark detection, face frontalization, and face recognition. Third, a feature transfer learning method is presented to solve the problem of insufficient intra-class variation via feature-level data augmentation. A Gaussian prior is assumed across all the regular classes and the variance are transferred from regular classes to long-tail classes. Further, an alternating training regimen is proposed to simultaneously achieve less biased decision boundaries and more discriminative representations. Extensive experiments have demonstrated the effectiveness of the proposed feature transfer framework.\"--Abstract.","abstract_html":"&quot;Face recognition has been advanced a lot in recent years thanks to the development of deep neural networks. The large intra-class variations in pose, illumination, and expression are the long-standing challenges. Learning a discriminative representation that is robust to these variations is the key. In the scenarios of profile pose or long-tail training data, image or feature-level data augmentation is needed. This dissertation presents three different methods to solve these problems. First, we explore a multi-task Convolutional Neural Network (CNN) that aims to leverage side tasks to improve representation learning. A pose-directed multi-task CNN is introduced to better handle pose variation. The proposed framework is effective in pose-invariant face recognition. Second, we propose a Face Frontalization-Generative Adversarial Network (FF-GAN) that can generate a frontal face even from an input image with extreme profile pose. FF-GAN handles pose variation from the perspective of image-level data augmentation. Multiple loss functions are proposed to achieve large-pose face frontalization. The proposed approach is evaluated on various tasks including face reconstruction, landmark detection, face frontalization, and face recognition. Third, a feature transfer learning method is presented to solve the problem of insufficient intra-class variation via feature-level data augmentation. A Gaussian prior is assumed across all the regular classes and the variance are transferred from regular classes to long-tail classes. Further, an alternating training regimen is proposed to simultaneously achieve less biased decision boundaries and more discriminative representations. Extensive experiments have demonstrated the effectiveness of the proposed feature transfer framework.&quot;--Abstract.","abstract_has_math":false,"creators":["Yin, Xi"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Liu, Xiaoming","Jain, Anil K.","Ross, Arun","Morris, Daniel D.","Michigan State University"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2018,"date_issued":"2018","date_published":"2018","updated_at":"2026-11-18T17:43:12Z","subjects":["Neural networks (Computer science)","Human face recognition (Computer science)"],"languages":["English"],"rights":["In Copyright"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:19404","isbn:9780438289192","isbn:0438289196","oclc:1078987923","umi:10930064","local:Yin_grad.msu_0128D_16300"],"render_values":[{"text":"etd:19404","href":null,"code":true},{"text":"isbn:9780438289192","href":null,"code":true},{"text":"isbn:0438289196","href":null,"code":true},{"text":"oclc:1078987923","href":null,"code":true},{"text":"umi:10930064","href":null,"code":true},{"text":"local:Yin_grad.msu_0128D_16300","href":null,"code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://doi.org/doi:10.25335/ag2x-4b04","outbound_label":"DOI","outbound_source":"dc:identifier","record_url":"https://doi.org/doi:10.25335/ag2x-4b04","record_label":"Repository record","record_kind":"doi","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://d.lib.msu.edu/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Ad.lib.msu.edu%3Aetd_19404","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Liu, Xiaoming","Jain, Anil K.","Ross, Arun","Morris, Daniel D.","Michigan State University"]},{"key":"dc:creator","label":"Author","values":["Yin, Xi"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2018"]},{"key":"dc:relation","label":"Dc Relation","values":["Electronic Theses & Dissertations"]},{"key":"dc:type","label":"Dc Type","values":["Text","Theses"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Neural networks (Computer science)","Human face recognition (Computer science)"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["English"]},{"key":"dc:rights","label":"Dc Rights","values":["In Copyright"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:19404","isbn:9780438289192","isbn:0438289196","oclc:1078987923","umi:10930064","local:Yin_grad.msu_0128D_16300","https://doi.org/doi:10.25335/ag2x-4b04"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["\"Face recognition has been advanced a lot in recent years thanks to the development of deep neural networks. The large intra-class variations in pose, illumination, and expression are the long-standing challenges. Learning a discriminative representation that is robust to these variations is the key. In the scenarios of profile pose or long-tail training data, image or feature-level data augmentation is needed. This dissertation presents three different methods to solve these problems. First, we explore a multi-task Convolutional Neural Network (CNN) that aims to leverage side tasks to improve representation learning. A pose-directed multi-task CNN is introduced to better handle pose variation. The proposed framework is effective in pose-invariant face recognition. Second, we propose a Face Frontalization-Generative Adversarial Network (FF-GAN) that can generate a frontal face even from an input image with extreme profile pose. FF-GAN handles pose variation from the perspective of image-level data augmentation. Multiple loss functions are proposed to achieve large-pose face frontalization. The proposed approach is evaluated on various tasks including face reconstruction, landmark detection, face frontalization, and face recognition. Third, a feature transfer learning method is presented to solve the problem of insufficient intra-class variation via feature-level data augmentation. A Gaussian prior is assumed across all the regular classes and the variance are transferred from regular classes to long-tail classes. Further, an alternating training regimen is proposed to simultaneously achieve less biased decision boundaries and more discriminative representations. Extensive experiments have demonstrated the effectiveness of the proposed feature transfer framework.\"--Abstract.","Thesis (Ph. D.)--Michigan State University. Computer Science, 2018","Includes bibliographical references (102-113)"]},{"key":"dc:format","label":"Dc Format","values":["xi, 113 pages","application/pdf"]},{"key":"dc:title","label":"Title","values":["Representation learning and image synthesis for deep face recognition"]}]}],"canonical_facts":{"dc:contributor":["Liu, Xiaoming","Jain, Anil K.","Ross, Arun","Morris, Daniel D.","Michigan State University"],"dc:creator":["Yin, Xi"],"dc:date":["2018"],"dc:description":["\"Face recognition has been advanced a lot in recent years thanks to the development of deep neural networks. The large intra-class variations in pose, illumination, and expression are the long-standing challenges. Learning a discriminative representation that is robust to these variations is the key. In the scenarios of profile pose or long-tail training data, image or feature-level data augmentation is needed. This dissertation presents three different methods to solve these problems. First, we explore a multi-task Convolutional Neural Network (CNN) that aims to leverage side tasks to improve representation learning. A pose-directed multi-task CNN is introduced to better handle pose variation. The proposed framework is effective in pose-invariant face recognition. Second, we propose a Face Frontalization-Generative Adversarial Network (FF-GAN) that can generate a frontal face even from an input image with extreme profile pose. FF-GAN handles pose variation from the perspective of image-level data augmentation. Multiple loss functions are proposed to achieve large-pose face frontalization. The proposed approach is evaluated on various tasks including face reconstruction, landmark detection, face frontalization, and face recognition. Third, a feature transfer learning method is presented to solve the problem of insufficient intra-class variation via feature-level data augmentation. A Gaussian prior is assumed across all the regular classes and the variance are transferred from regular classes to long-tail classes. Further, an alternating training regimen is proposed to simultaneously achieve less biased decision boundaries and more discriminative representations. Extensive experiments have demonstrated the effectiveness of the proposed feature transfer framework.\"--Abstract.","Thesis (Ph. D.)--Michigan State University. Computer Science, 2018","Includes bibliographical references (102-113)"],"dc:format":["xi, 113 pages","application/pdf"],"dc:identifier":["etd:19404","isbn:9780438289192","isbn:0438289196","oclc:1078987923","umi:10930064","local:Yin_grad.msu_0128D_16300","https://doi.org/doi:10.25335/ag2x-4b04"],"dc:language":["English"],"dc:relation":["Electronic Theses & Dissertations"],"dc:rights":["In Copyright"],"dc:subject":["Neural networks (Computer science)","Human face recognition (Computer science)"],"dc:title":["Representation learning and image synthesis for deep face recognition"],"dc:type":["Text","Theses"]},"updated_at":"2026-11-18T17:43:12Z"}