{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/105694"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/105694","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Application of capsule networks for image classification on complex datasets","abstract":"Capsule Network, introduced in 2017 by Sabour, Hinton, and Frost, has sparked great interest in the computer vision and deep learning community and offers a paradigm shift in neural computation. In CapsNet, Sabour et. al. replace classical notions of scalar neural computation with a vectorised approach. This allows CapsNet to describe input images not only by the presence of constituent features but also by the pose of detected features, thus imparting view-point and pose invariance. Hinton’s group and the research community at large have applied CapsNets to a number of specific problems and achieved state-of-the-art performance. In contrast, this thesis studies CapsNet by applying it to complex real world datasets like CIFAR10 and CIFAR100 where the CapsNet’s performance is still unproven. We investigate the operational characteristics of CapsNet for the CIFAR10 problem and identify several practical limitations of Capsules that inhibit their performance in an industrial setting. The contribution of this research is the introduction of residual blocks of primary capsule layers. We developed a novel architecture for CIFAR10 classification, called ResCapsNet, and find that the model increases validation accuracy to 78.54% from 71.04% achieved by the baseline CapsNet, at the marginal cost of increasing the number of parameters from 22 million to 25 million. In addition, to extend the generalization of capsules into deeper networks, we discuss the application of Capsules as hidden layers in CIFAR100 classification and show that Capsules are largely ineffective in a latent unsupervised setting. For active supervision of hidden capsules, we propose methods to train hidden capsules as super-class detectors prior to final classification.","abstract_html":"Capsule Network, introduced in 2017 by Sabour, Hinton, and Frost, has sparked great interest in the computer vision and deep learning community and offers a paradigm shift in neural computation. In CapsNet, Sabour et. al. replace classical notions of scalar neural computation with a vectorised approach. This allows CapsNet to describe input images not only by the presence of constituent features but also by the pose of detected features, thus imparting view-point and pose invariance. Hinton’s group and the research community at large have applied CapsNets to a number of specific problems and achieved state-of-the-art performance. In contrast, this thesis studies CapsNet by applying it to complex real world datasets like CIFAR10 and CIFAR100 where the CapsNet’s performance is still unproven. We investigate the operational characteristics of CapsNet for the CIFAR10 problem and identify several practical limitations of Capsules that inhibit their performance in an industrial setting. The contribution of this research is the introduction of residual blocks of primary capsule layers. We developed a novel architecture for CIFAR10 classification, called ResCapsNet, and find that the model increases validation accuracy to 78.54% from 71.04% achieved by the baseline CapsNet, at the marginal cost of increasing the number of parameters from 22 million to 25 million. In addition, to extend the generalization of capsules into deeper networks, we discuss the application of Capsules as hidden layers in CIFAR100 classification and show that Capsules are largely ineffective in a latent unsupervised setting. For active supervision of hidden capsules, we propose methods to train hidden capsules as super-class detectors prior to final classification.","abstract_has_math":false,"creators":["Goswami, Deborshi"],"institution":"University of Illinois at Urbana-Champaign","degree_name":"M.S.","degree_level":"Thesis","degree_discipline":"Industrial Engineering","degree_department":null,"school":null,"contributors":["Sun, Ruoyu"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2019,"date_issued":"2019-11-26T20:35:09Z","date_published":"2019-11-26T20:35:09Z","updated_at":"2026-07-22T22:24:44Z","subjects":["Capsule Networks, Image Classification, Dynamic Routing, Residual Networks"],"languages":["en"],"rights":["Copyright 2019 Deborshi Goswami"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/2142/105694","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Sun, Ruoyu"]},{"key":"dc:creator","label":"Author","values":["Goswami, Deborshi"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2019-11-26T20:35:09Z","2019-07-15","2019-08"]},{"key":"dc:type","label":"Dc Type","values":["text"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Industrial Engineering"]},{"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":["Capsule Networks, Image Classification, Dynamic Routing, Residual Networks"]}]},{"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 Deborshi Goswami"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/2142/105694"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Capsule Network, introduced in 2017 by Sabour, Hinton, and Frost, has sparked great interest in the computer vision and deep learning community and offers a paradigm shift in neural computation. In CapsNet, Sabour et. al. replace classical notions of scalar neural computation with a vectorised approach. This allows CapsNet to describe input images not only by the presence of constituent features but also by the pose of detected features, thus imparting view-point and pose invariance. Hinton’s group and the research community at large have applied CapsNets to a number of specific problems and achieved state-of-the-art performance. In contrast, this thesis studies CapsNet by applying it to complex real world datasets like CIFAR10 and CIFAR100 where the CapsNet’s performance is still unproven. We investigate the operational characteristics of CapsNet for the CIFAR10 problem and identify several practical limitations of Capsules that inhibit their performance in an industrial setting. The contribution of this research is the introduction of residual blocks of primary capsule layers. We developed a novel architecture for CIFAR10 classification, called ResCapsNet, and find that the model increases validation accuracy to 78.54% from 71.04% achieved by the baseline CapsNet, at the marginal cost of increasing the number of parameters from 22 million to 25 million. In addition, to extend the generalization of capsules into deeper networks, we discuss the application of Capsules as hidden layers in CIFAR100 classification and show that Capsules are largely ineffective in a latent unsupervised setting. For active supervision of hidden capsules, we propose methods to train hidden capsules as super-class detectors prior to final classification.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2019-11-26 without embargo terms","The student, Deborshi Goswami, accepted the attached license on 2019-07-12 at 15:31.","The student, Deborshi Goswami, submitted this Thesis for approval on 2019-07-12 at 15:40.","This Thesis was approved for publication on 2019-07-15 at 16:53.","DSpace SAF Submission Ingestion Package generated from Vireo submission #14309 on 2019-11-26 at 12:53:38","Made available in DSpace on 2019-11-26T20:35:09Z (GMT). No. of bitstreams: 2 GOSWAMI-THESIS-2019.pdf: 487668 bytes, checksum: 5176ff7bdef1f6a22fe61c227ca78b0e (MD5) LICENSE.txt: 4213 bytes, checksum: ce3296b9481d31e7ed69b3a116619634 (MD5) Previous issue date: 2019-07-15"]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Application of capsule networks for image classification on complex datasets"]}]}],"canonical_facts":{"dc:contributor":["Sun, Ruoyu"],"dc:creator":["Goswami, Deborshi"],"dc:date":["2019-11-26T20:35:09Z","2019-07-15","2019-08"],"dc:description":["Capsule Network, introduced in 2017 by Sabour, Hinton, and Frost, has sparked great interest in the computer vision and deep learning community and offers a paradigm shift in neural computation. In CapsNet, Sabour et. al. replace classical notions of scalar neural computation with a vectorised approach. This allows CapsNet to describe input images not only by the presence of constituent features but also by the pose of detected features, thus imparting view-point and pose invariance. Hinton’s group and the research community at large have applied CapsNets to a number of specific problems and achieved state-of-the-art performance. In contrast, this thesis studies CapsNet by applying it to complex real world datasets like CIFAR10 and CIFAR100 where the CapsNet’s performance is still unproven. We investigate the operational characteristics of CapsNet for the CIFAR10 problem and identify several practical limitations of Capsules that inhibit their performance in an industrial setting. The contribution of this research is the introduction of residual blocks of primary capsule layers. We developed a novel architecture for CIFAR10 classification, called ResCapsNet, and find that the model increases validation accuracy to 78.54% from 71.04% achieved by the baseline CapsNet, at the marginal cost of increasing the number of parameters from 22 million to 25 million. In addition, to extend the generalization of capsules into deeper networks, we discuss the application of Capsules as hidden layers in CIFAR100 classification and show that Capsules are largely ineffective in a latent unsupervised setting. For active supervision of hidden capsules, we propose methods to train hidden capsules as super-class detectors prior to final classification.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2019-11-26 without embargo terms","The student, Deborshi Goswami, accepted the attached license on 2019-07-12 at 15:31.","The student, Deborshi Goswami, submitted this Thesis for approval on 2019-07-12 at 15:40.","This Thesis was approved for publication on 2019-07-15 at 16:53.","DSpace SAF Submission Ingestion Package generated from Vireo submission #14309 on 2019-11-26 at 12:53:38","Made available in DSpace on 2019-11-26T20:35:09Z (GMT). No. of bitstreams: 2 GOSWAMI-THESIS-2019.pdf: 487668 bytes, checksum: 5176ff7bdef1f6a22fe61c227ca78b0e (MD5) LICENSE.txt: 4213 bytes, checksum: ce3296b9481d31e7ed69b3a116619634 (MD5) Previous issue date: 2019-07-15"],"dc:format":["application/pdf"],"dc:identifier":["http://hdl.handle.net/2142/105694"],"dc:language":["en"],"dc:rights":["Copyright 2019 Deborshi Goswami"],"dc:subject":["Capsule Networks, Image Classification, Dynamic Routing, Residual Networks"],"dc:title":["Application of capsule networks for image classification on complex datasets"],"dc:type":["text"],"thesis:degree_discipline":["Industrial Engineering"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["M.S."],"thesis:institution_name":["University of Illinois at Urbana-Champaign"]},"updated_at":"2026-07-22T22:24:44Z"}