{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/104903"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/104903","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Three ploys for robust co-generation with generative adversarial nets","abstract":"Generative adversarial nets (GANs) and variational auto-encoders enable accurate modeling of high-dimensional data distributions by forward propagating a sample drawn from a latent space. However, an often overlooked shortcoming is their inability to find an arbitrary marginal distribution, which is useful for completion of missing data in tasks like super-resolution, image inpainting, etc., where we don’t know the missing part ahead of time. To address such applications it seems intuitive at first to search for that latent space sample which ‘best’ matches the observations. However, irrespective of the GAN loss, unexpected challenges arise: we find that the energy landscape of well trained generators is extremely hard to optimize, exhibiting ‘folds’ that are very hard to overcome. To address this issue, in this thesis, three ploys are proposed which help to address the challenge for all investigated GAN losses and which yield more accurate reconstructions, quantitatively and qualitatively.","abstract_html":"Generative adversarial nets (GANs) and variational auto-encoders enable accurate modeling of high-dimensional data distributions by forward propagating a sample drawn from a latent space. However, an often overlooked shortcoming is their inability to find an arbitrary marginal distribution, which is useful for completion of missing data in tasks like super-resolution, image inpainting, etc., where we don’t know the missing part ahead of time. To address such applications it seems intuitive at first to search for that latent space sample which ‘best’ matches the observations. However, irrespective of the GAN loss, unexpected challenges arise: we find that the energy landscape of well trained generators is extremely hard to optimize, exhibiting ‘folds’ that are very hard to overcome. To address this issue, in this thesis, three ploys are proposed which help to address the challenge for all investigated GAN losses and which yield more accurate reconstructions, quantitatively and qualitatively.","abstract_has_math":false,"creators":["Kothapalli, Krishna Harsha Reddy"],"institution":"University of Illinois at Urbana-Champaign","degree_name":"M.S.","degree_level":"Thesis","degree_discipline":"Computer Science","degree_department":null,"school":null,"contributors":["Schwing, Alexander"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2019,"date_issued":"2019-08-23T20:00:14Z","date_published":"2019-08-23T20:00:14Z","updated_at":"2026-07-22T22:24:42Z","subjects":["Deep Learning","Machine Learning","GAN","Cogeneration"],"languages":["en"],"rights":["Copyright 2019 Krishna Harsha Reddy Kothapalli"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/2142/104903","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Schwing, Alexander"]},{"key":"dc:creator","label":"Author","values":["Kothapalli, Krishna Harsha Reddy"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2019-08-23T20:00:14Z","2019-04-23","2019-05"]},{"key":"dc:type","label":"Dc Type","values":["text"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["M.S."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Illinois at Urbana-Champaign"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Deep Learning","Machine Learning","GAN","Cogeneration"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright 2019 Krishna Harsha Reddy Kothapalli"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/2142/104903"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Generative adversarial nets (GANs) and variational auto-encoders enable accurate modeling of high-dimensional data distributions by forward propagating a sample drawn from a latent space. However, an often overlooked shortcoming is their inability to find an arbitrary marginal distribution, which is useful for completion of missing data in tasks like super-resolution, image inpainting, etc., where we don’t know the missing part ahead of time. To address such applications it seems intuitive at first to search for that latent space sample which ‘best’ matches the observations. However, irrespective of the GAN loss, unexpected challenges arise: we find that the energy landscape of well trained generators is extremely hard to optimize, exhibiting ‘folds’ that are very hard to overcome. To address this issue, in this thesis, three ploys are proposed which help to address the challenge for all investigated GAN losses and which yield more accurate reconstructions, quantitatively and qualitatively.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2019-08-22 without embargo terms","The student, Krishna Harsha Reddy Kothapalli, accepted the attached license on 2019-04-22 at 16:08.","The student, Krishna Harsha Reddy Kothapalli, submitted this Thesis for approval on 2019-04-22 at 16:29.","This Thesis was approved for publication on 2019-04-23 at 08:23.","DSpace SAF Submission Ingestion Package generated from Vireo submission #13829 on 2019-08-22 at 14:45:39","Made available in DSpace on 2019-08-23T20:00:14Z (GMT). No. of bitstreams: 3 KOTHAPALLI-THESIS-2019.pdf: 41161785 bytes, checksum: 1f83eff9abc02ee91aee94a662074921 (MD5) cs_thesis_latex.zip: 142770934 bytes, checksum: 5b4468ab70282bc9d2ed4698ddf78209 (MD5) LICENSE.txt: 4228 bytes, checksum: 9a27e1c379f1336b4248e6d9d974d731 (MD5) Previous issue date: 2019-04-23"]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Three ploys for robust co-generation with generative adversarial nets"]}]}],"canonical_facts":{"dc:contributor":["Schwing, Alexander"],"dc:creator":["Kothapalli, Krishna Harsha Reddy"],"dc:date":["2019-08-23T20:00:14Z","2019-04-23","2019-05"],"dc:description":["Generative adversarial nets (GANs) and variational auto-encoders enable accurate modeling of high-dimensional data distributions by forward propagating a sample drawn from a latent space. However, an often overlooked shortcoming is their inability to find an arbitrary marginal distribution, which is useful for completion of missing data in tasks like super-resolution, image inpainting, etc., where we don’t know the missing part ahead of time. To address such applications it seems intuitive at first to search for that latent space sample which ‘best’ matches the observations. However, irrespective of the GAN loss, unexpected challenges arise: we find that the energy landscape of well trained generators is extremely hard to optimize, exhibiting ‘folds’ that are very hard to overcome. To address this issue, in this thesis, three ploys are proposed which help to address the challenge for all investigated GAN losses and which yield more accurate reconstructions, quantitatively and qualitatively.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2019-08-22 without embargo terms","The student, Krishna Harsha Reddy Kothapalli, accepted the attached license on 2019-04-22 at 16:08.","The student, Krishna Harsha Reddy Kothapalli, submitted this Thesis for approval on 2019-04-22 at 16:29.","This Thesis was approved for publication on 2019-04-23 at 08:23.","DSpace SAF Submission Ingestion Package generated from Vireo submission #13829 on 2019-08-22 at 14:45:39","Made available in DSpace on 2019-08-23T20:00:14Z (GMT). No. of bitstreams: 3 KOTHAPALLI-THESIS-2019.pdf: 41161785 bytes, checksum: 1f83eff9abc02ee91aee94a662074921 (MD5) cs_thesis_latex.zip: 142770934 bytes, checksum: 5b4468ab70282bc9d2ed4698ddf78209 (MD5) LICENSE.txt: 4228 bytes, checksum: 9a27e1c379f1336b4248e6d9d974d731 (MD5) Previous issue date: 2019-04-23"],"dc:format":["application/pdf"],"dc:identifier":["http://hdl.handle.net/2142/104903"],"dc:language":["en"],"dc:rights":["Copyright 2019 Krishna Harsha Reddy Kothapalli"],"dc:subject":["Deep Learning","Machine Learning","GAN","Cogeneration"],"dc:title":["Three ploys for robust co-generation with generative adversarial nets"],"dc:type":["text"],"thesis:degree_discipline":["Computer Science"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["M.S."],"thesis:institution_name":["University of Illinois at Urbana-Champaign"]},"updated_at":"2026-07-22T22:24:42Z"}