{"id":{"repo_id":"trento","oai_identifier":"oai:iris.unitn.it:11572/458913"},"canonical_url":"https://search.dev.ndltd.org/etd/trento/oai:iris.unitn.it:11572/458913","repository":{"repo_id":"trento","name":"Università degli Studi di Trento","base_url":"https://iris.unitn.it/oai/request"},"display":{"title":"Learning without Labels - Reducing Supervision in Training, Inference, and Evaluation of Deep Neural Networks","abstract":"This thesis investigates how the reliance on supervision can be reduced across the entire deep learning pipeline. In the training phase, we explore unsupervised fine-tuning, focusing on Source-Free Unsupervised Domain Adaptation scenarios in visual tasks such as Facial Expression Recognition and video-based Action Recognition, primarily leveraging self-supervision and self-training. At inference, we address the challenge of removing fixed output vocabularies from Vision Language Models by formalizing the tasks of Vocabulary-free Image Classification and Vocabulary-free Semantic Segmentation and by introducing a family of efficient methods that adapt CLIP to the tasks. We also evaluate Large Multimodal Models under a similar constrained scenario, analyzing their predictions, categorizing their mistakes, and proposing tailored solutions to optimize their performance. Finally, we investigate unsupervised evaluation by proposing a framework that uses a Large Language Model and modular tools to automatically generate, execute, and interpret evaluation experiments for Large Multimodal Models without ground-truth labels. By reducing the need for human supervision at every stage of the deep learning pipeline, this thesis contributes toward a more flexible and efficient paradigm for developing and deploying deep neural networks in real-world, data-scarce, and open-ended settings.","abstract_html":"This thesis investigates how the reliance on supervision can be reduced across the entire deep learning pipeline. In the training phase, we explore unsupervised fine-tuning, focusing on Source-Free Unsupervised Domain Adaptation scenarios in visual tasks such as Facial Expression Recognition and video-based Action Recognition, primarily leveraging self-supervision and self-training. At inference, we address the challenge of removing fixed output vocabularies from Vision Language Models by formalizing the tasks of Vocabulary-free Image Classification and Vocabulary-free Semantic Segmentation and by introducing a family of efficient methods that adapt CLIP to the tasks. We also evaluate Large Multimodal Models under a similar constrained scenario, analyzing their predictions, categorizing their mistakes, and proposing tailored solutions to optimize their performance. Finally, we investigate unsupervised evaluation by proposing a framework that uses a Large Language Model and modular tools to automatically generate, execute, and interpret evaluation experiments for Large Multimodal Models without ground-truth labels. By reducing the need for human supervision at every stage of the deep learning pipeline, this thesis contributes toward a more flexible and efficient paradigm for developing and deploying deep neural networks in real-world, data-scarce, and open-ended settings.","abstract_has_math":false,"creators":["Conti, Alessandro"],"institution":"Università degli studi di Trento","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Ricci, Elisa","Rota, Paolo"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-07-17","date_published":"2025-07-17","updated_at":"2026-07-24T05:04:45Z","subjects":["Fine-tuning with domain shift, Inference without labels, Automatic benchmarking"],"languages":["eng"],"rights":["info:eu-repo/semantics/openAccess","license:Creative commons","license uri:http://creativecommons.org/licenses/by-nc-nd/4.0/"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["http://dx.doi.org/10.15168/11572_458913","10.15168/11572_458913"],"render_values":[{"text":"http://dx.doi.org/10.15168/11572_458913","href":"http://dx.doi.org/10.15168/11572_458913","code":true},{"text":"10.15168/11572_458913","href":"https://doi.org/10.15168/11572_458913","code":true}]}]},"links":{"outbound_url":"https://hdl.handle.net/11572/458913","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Conti, Alessandro","Ricci, Elisa","Rota, Paolo"]},{"key":"dc:creator","label":"Author","values":["Conti, Alessandro"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2025-07-17"]},{"key":"dc:publisher","label":"Institution","values":["Università degli studi di Trento","place:TRENTO"]},{"key":"dc:relation","label":"Dc Relation","values":["firstpage:1","lastpage:195","numberofpages:195"]},{"key":"dc:type","label":"Dc Type","values":["info:eu-repo/semantics/doctoralThesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Fine-tuning with domain shift, Inference without labels, Automatic benchmarking"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["info:eu-repo/semantics/openAccess","license:Creative commons","license uri:http://creativecommons.org/licenses/by-nc-nd/4.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://hdl.handle.net/11572/458913","http://dx.doi.org/10.15168/11572_458913","10.15168/11572_458913"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["This thesis investigates how the reliance on supervision can be reduced across the entire deep learning pipeline. In the training phase, we explore unsupervised fine-tuning, focusing on Source-Free Unsupervised Domain Adaptation scenarios in visual tasks such as Facial Expression Recognition and video-based Action Recognition, primarily leveraging self-supervision and self-training. At inference, we address the challenge of removing fixed output vocabularies from Vision Language Models by formalizing the tasks of Vocabulary-free Image Classification and Vocabulary-free Semantic Segmentation and by introducing a family of efficient methods that adapt CLIP to the tasks. We also evaluate Large Multimodal Models under a similar constrained scenario, analyzing their predictions, categorizing their mistakes, and proposing tailored solutions to optimize their performance. Finally, we investigate unsupervised evaluation by proposing a framework that uses a Large Language Model and modular tools to automatically generate, execute, and interpret evaluation experiments for Large Multimodal Models without ground-truth labels. By reducing the need for human supervision at every stage of the deep learning pipeline, this thesis contributes toward a more flexible and efficient paradigm for developing and deploying deep neural networks in real-world, data-scarce, and open-ended settings."]},{"key":"dc:title","label":"Title","values":["Learning without Labels - Reducing Supervision in Training, Inference, and Evaluation of Deep Neural Networks"]}]}],"canonical_facts":{"dc:contributor":["Conti, Alessandro","Ricci, Elisa","Rota, Paolo"],"dc:creator":["Conti, Alessandro"],"dc:date":["2025-07-17"],"dc:description":["This thesis investigates how the reliance on supervision can be reduced across the entire deep learning pipeline. In the training phase, we explore unsupervised fine-tuning, focusing on Source-Free Unsupervised Domain Adaptation scenarios in visual tasks such as Facial Expression Recognition and video-based Action Recognition, primarily leveraging self-supervision and self-training. At inference, we address the challenge of removing fixed output vocabularies from Vision Language Models by formalizing the tasks of Vocabulary-free Image Classification and Vocabulary-free Semantic Segmentation and by introducing a family of efficient methods that adapt CLIP to the tasks. We also evaluate Large Multimodal Models under a similar constrained scenario, analyzing their predictions, categorizing their mistakes, and proposing tailored solutions to optimize their performance. Finally, we investigate unsupervised evaluation by proposing a framework that uses a Large Language Model and modular tools to automatically generate, execute, and interpret evaluation experiments for Large Multimodal Models without ground-truth labels. By reducing the need for human supervision at every stage of the deep learning pipeline, this thesis contributes toward a more flexible and efficient paradigm for developing and deploying deep neural networks in real-world, data-scarce, and open-ended settings."],"dc:identifier":["https://hdl.handle.net/11572/458913","http://dx.doi.org/10.15168/11572_458913","10.15168/11572_458913"],"dc:language":["eng"],"dc:publisher":["Università degli studi di Trento","place:TRENTO"],"dc:relation":["firstpage:1","lastpage:195","numberofpages:195"],"dc:rights":["info:eu-repo/semantics/openAccess","license:Creative commons","license uri:http://creativecommons.org/licenses/by-nc-nd/4.0/"],"dc:subject":["Fine-tuning with domain shift, Inference without labels, Automatic benchmarking"],"dc:title":["Learning without Labels - Reducing Supervision in Training, Inference, and Evaluation of Deep Neural Networks"],"dc:type":["info:eu-repo/semantics/doctoralThesis"]},"updated_at":"2026-07-24T05:04:45Z"}