{"id":{"repo_id":"waikato-masters","oai_identifier":"oai:researchcommons.waikato.ac.nz:10289/16548"},"canonical_url":"https://search.dev.ndltd.org/etd/waikato-masters/oai:researchcommons.waikato.ac.nz:10289/16548","repository":{"repo_id":"waikato-masters","name":"University Waikato","base_url":"https://researchcommons.waikato.ac.nz/server/oai/request"},"display":{"title":"Feature extractor stacking for cross-domain few-shot learning","abstract":"Cross-domain few-shot learning (CDFSL) addresses learning problems where knowledge needs to be transferred from one or more source domains into an instance-scarce target domain with an explicitly different distribution. Recently published CDFSL methods generally construct a universal model that combines knowledge of multiple source domains into one feature extractor. This enables efficient inference but necessitates re-computation of the extractor whenever a new source domain is added. Some of these methods are also incompatible with heterogeneous source domain extractor architectures. The first part of this thesis proposes feature extractor stacking (FES), a new CDFSL method for combining information from a collection of extractors, that can utilise heterogeneous pretrained extractors out of the box and does not maintain a universal model that needs to be re-computed when its extractor collection is updated. We present the basic FES algorithm, which is inspired by the classic stacked generalisation approach, and also introduce two variants: convolutional FES (ConFES) and regularised FES (ReFES). Given a target-domain task, these algorithms fine-tune each extractor independently, use cross-validation to extract training data for stacked generalisation from the support set, and learn a simple linear stacking classifier from this data. We evaluate our FES methods on the well-known Meta-Dataset benchmark, targeting image classification with convolutional neural networks, and show that they can achieve state-of-the-art performance. The second part of this thesis proposes an efficient semi-supervised learning method that applies self-training to the classification head only and show that it can yield very consistent improvements in average performance in the Meta-Dataset benchmark for cross-domain few-shot learning when applied with FES and other contemporary methods utilising centroid-based classification. The third part of this thesis proposes a bidirectional snapshot selection strategy for FES, leveraging its cross-validation process and the ordered nature of its snapshots, and demonstrates that a 95% snapshot reduction can be achieved while retaining the same level of accuracy.","abstract_html":"Cross-domain few-shot learning (CDFSL) addresses learning problems where knowledge needs to be transferred from one or more source domains into an instance-scarce target domain with an explicitly different distribution. Recently published CDFSL methods generally construct a universal model that combines knowledge of multiple source domains into one feature extractor. This enables efficient inference but necessitates re-computation of the extractor whenever a new source domain is added. Some of these methods are also incompatible with heterogeneous source domain extractor architectures. The first part of this thesis proposes feature extractor stacking (FES), a new CDFSL method for combining information from a collection of extractors, that can utilise heterogeneous pretrained extractors out of the box and does not maintain a universal model that needs to be re-computed when its extractor collection is updated. We present the basic FES algorithm, which is inspired by the classic stacked generalisation approach, and also introduce two variants: convolutional FES (ConFES) and regularised FES (ReFES). Given a target-domain task, these algorithms fine-tune each extractor independently, use cross-validation to extract training data for stacked generalisation from the support set, and learn a simple linear stacking classifier from this data. We evaluate our FES methods on the well-known Meta-Dataset benchmark, targeting image classification with convolutional neural networks, and show that they can achieve state-of-the-art performance. The second part of this thesis proposes an efficient semi-supervised learning method that applies self-training to the classification head only and show that it can yield very consistent improvements in average performance in the Meta-Dataset benchmark for cross-domain few-shot learning when applied with FES and other contemporary methods utilising centroid-based classification. The third part of this thesis proposes a bidirectional snapshot selection strategy for FES, leveraging its cross-validation process and the ordered nature of its snapshots, and demonstrates that a 95% snapshot reduction can be achieved while retaining the same level of accuracy.","abstract_has_math":false,"creators":["Wang, Hongyu"],"institution":"The University of Waikato","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Frank, Eibe","Pfahringer, Bernhard","Holmes, Geoffrey"],"committee_chairs":[],"committee_members":[],"year":2024,"date_issued":"2024","date_published":"2024","updated_at":"2026-07-24T05:57:38Z","subjects":["cross-domain few-shot learning","pretrained feature extractors","stacking","semi-supervised learning","self-training","feature subset selection"],"languages":[],"rights":["All items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated."],"rights_urls":["https://researchcommons.waikato.ac.nz/bitstreams/5cbcf952-0f23-4d21-be3a-007c0f393ccb/download"],"identifier_entries":[]},"links":{"outbound_url":null,"outbound_label":null,"outbound_source":null},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Frank, Eibe","Pfahringer, Bernhard","Holmes, Geoffrey"]},{"key":"dc:creator","label":"Author","values":["Wang, Hongyu"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.issued","label":"Date","values":["2024"]},{"key":"dc:publisher.institution","label":"Dc Publisher Institution","values":["The University of Waikato"]},{"key":"dc:relation.isreferencedby","label":"Dc Relation Isreferencedby","values":["https://hdl.handle.net/10289/16548"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["cross-domain few-shot learning","pretrained feature extractors","stacking","semi-supervised learning","self-training","feature subset selection"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:rights","label":"Dc Rights","values":["https://researchcommons.waikato.ac.nz/bitstreams/5cbcf952-0f23-4d21-be3a-007c0f393ccb/download","All items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated."]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://researchcommons.waikato.ac.nz/bitstreams/bcf8d8c5-80fe-4729-8b54-21f32afa085c/download"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Cross-domain few-shot learning (CDFSL) addresses learning problems where knowledge needs to be transferred from one or more source domains into an instance-scarce target domain with an explicitly different distribution. Recently published CDFSL methods generally construct a universal model that combines knowledge of multiple source domains into one feature extractor. This enables efficient inference but necessitates re-computation of the extractor whenever a new source domain is added. Some of these methods are also incompatible with heterogeneous source domain extractor architectures. The first part of this thesis proposes feature extractor stacking (FES), a new CDFSL method for combining information from a collection of extractors, that can utilise heterogeneous pretrained extractors out of the box and does not maintain a universal model that needs to be re-computed when its extractor collection is updated. We present the basic FES algorithm, which is inspired by the classic stacked generalisation approach, and also introduce two variants: convolutional FES (ConFES) and regularised FES (ReFES). Given a target-domain task, these algorithms fine-tune each extractor independently, use cross-validation to extract training data for stacked generalisation from the support set, and learn a simple linear stacking classifier from this data. We evaluate our FES methods on the well-known Meta-Dataset benchmark, targeting image classification with convolutional neural networks, and show that they can achieve state-of-the-art performance. The second part of this thesis proposes an efficient semi-supervised learning method that applies self-training to the classification head only and show that it can yield very consistent improvements in average performance in the Meta-Dataset benchmark for cross-domain few-shot learning when applied with FES and other contemporary methods utilising centroid-based classification. The third part of this thesis proposes a bidirectional snapshot selection strategy for FES, leveraging its cross-validation process and the ordered nature of its snapshots, and demonstrates that a 95% snapshot reduction can be achieved while retaining the same level of accuracy."]},{"key":"dc:format.checksum.md5","label":"Dc Format Checksum Md5","values":["e41938d7b788d6f17b1a67e0c4602aa0","e14202ab27e47ddb00d33097327ba050","7c7136817733234d89db35252d5212c7"]},{"key":"dc:title","label":"Title","values":["Feature extractor stacking for cross-domain few-shot learning"]}]}],"canonical_facts":{"dc:contributor.advisor":["Frank, Eibe","Pfahringer, Bernhard","Holmes, Geoffrey"],"dc:creator":["Wang, Hongyu"],"dc:date.issued":["2024"],"dc:description.abstract":["Cross-domain few-shot learning (CDFSL) addresses learning problems where knowledge needs to be transferred from one or more source domains into an instance-scarce target domain with an explicitly different distribution. Recently published CDFSL methods generally construct a universal model that combines knowledge of multiple source domains into one feature extractor. This enables efficient inference but necessitates re-computation of the extractor whenever a new source domain is added. Some of these methods are also incompatible with heterogeneous source domain extractor architectures. The first part of this thesis proposes feature extractor stacking (FES), a new CDFSL method for combining information from a collection of extractors, that can utilise heterogeneous pretrained extractors out of the box and does not maintain a universal model that needs to be re-computed when its extractor collection is updated. We present the basic FES algorithm, which is inspired by the classic stacked generalisation approach, and also introduce two variants: convolutional FES (ConFES) and regularised FES (ReFES). Given a target-domain task, these algorithms fine-tune each extractor independently, use cross-validation to extract training data for stacked generalisation from the support set, and learn a simple linear stacking classifier from this data. We evaluate our FES methods on the well-known Meta-Dataset benchmark, targeting image classification with convolutional neural networks, and show that they can achieve state-of-the-art performance. The second part of this thesis proposes an efficient semi-supervised learning method that applies self-training to the classification head only and show that it can yield very consistent improvements in average performance in the Meta-Dataset benchmark for cross-domain few-shot learning when applied with FES and other contemporary methods utilising centroid-based classification. The third part of this thesis proposes a bidirectional snapshot selection strategy for FES, leveraging its cross-validation process and the ordered nature of its snapshots, and demonstrates that a 95% snapshot reduction can be achieved while retaining the same level of accuracy."],"dc:format.checksum.md5":["e41938d7b788d6f17b1a67e0c4602aa0","e14202ab27e47ddb00d33097327ba050","7c7136817733234d89db35252d5212c7"],"dc:identifier.uri":["https://researchcommons.waikato.ac.nz/bitstreams/bcf8d8c5-80fe-4729-8b54-21f32afa085c/download"],"dc:publisher.institution":["The University of Waikato"],"dc:relation.isreferencedby":["https://hdl.handle.net/10289/16548"],"dc:rights":["https://researchcommons.waikato.ac.nz/bitstreams/5cbcf952-0f23-4d21-be3a-007c0f393ccb/download","All items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated."],"dc:subject":["cross-domain few-shot learning","pretrained feature extractors","stacking","semi-supervised learning","self-training","feature subset selection"],"dc:title":["Feature extractor stacking for cross-domain few-shot learning"],"dc:type":["Thesis"]},"updated_at":"2026-07-24T05:57:38Z"}