United We Learn Better: Harvesting Learning Improvements From Class Hierarchies Across Tasks

Attempts of learning from hierarchical taxonomies in computer vision have been mostly focusing on image classification. Though ways of best harvesting learning improvements from hierarchies in classification are far from being solved, there is a need to target these problems in other vision tasks su...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2021-07
Hauptverfasser: Sindi Shkodrani, Wang, Yu, Manfredi, Marco, Baka, Nóra
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Sindi Shkodrani
Wang, Yu
Manfredi, Marco
Baka, Nóra
description Attempts of learning from hierarchical taxonomies in computer vision have been mostly focusing on image classification. Though ways of best harvesting learning improvements from hierarchies in classification are far from being solved, there is a need to target these problems in other vision tasks such as object detection. As progress on the classification side is often dependent on hierarchical cross-entropy losses, novel detection architectures using sigmoid as an output function instead of softmax cannot easily apply these advances, requiring novel methods in detection. In this work we establish a theoretical framework based on probability and set theory for extracting parent predictions and a hierarchical loss that can be used across tasks, showing results across classification and detection benchmarks and opening up the possibility of hierarchical learning for sigmoid-based detection architectures.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2556541066</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2556541066</sourcerecordid><originalsourceid>FETCH-proquest_journals_25565410663</originalsourceid><addsrcrecordid>eNqNjs0KwjAQhIMgWLTvEPBcSNMfxZsWSwWPihehhLpqa5vobtrnN6IP4GmG-YZhRsyTURQGy1jKCfOJGiGETBcySSKPnY-6tnDhJ-B7UKj5BqwFXPFC4QBka337go_ZdU80A3SgLfEcTcezVhHxogZUWN1rIL6u0LjooOhBMza-qpbA_-mUzfPtISsCN_Pq3XrZmB61Q6V7kyZxKNI0-q_1Bnh3Qv4</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2556541066</pqid></control><display><type>article</type><title>United We Learn Better: Harvesting Learning Improvements From Class Hierarchies Across Tasks</title><source>Free E- Journals</source><creator>Sindi Shkodrani ; Wang, Yu ; Manfredi, Marco ; Baka, Nóra</creator><creatorcontrib>Sindi Shkodrani ; Wang, Yu ; Manfredi, Marco ; Baka, Nóra</creatorcontrib><description>Attempts of learning from hierarchical taxonomies in computer vision have been mostly focusing on image classification. Though ways of best harvesting learning improvements from hierarchies in classification are far from being solved, there is a need to target these problems in other vision tasks such as object detection. As progress on the classification side is often dependent on hierarchical cross-entropy losses, novel detection architectures using sigmoid as an output function instead of softmax cannot easily apply these advances, requiring novel methods in detection. In this work we establish a theoretical framework based on probability and set theory for extracting parent predictions and a hierarchical loss that can be used across tasks, showing results across classification and detection benchmarks and opening up the possibility of hierarchical learning for sigmoid-based detection architectures.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Classification ; Computer vision ; Entropy (Information theory) ; Hierarchies ; Image classification ; Learning ; Object recognition ; Set theory ; Taxonomy</subject><ispartof>arXiv.org, 2021-07</ispartof><rights>2021. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>780,784</link.rule.ids></links><search><creatorcontrib>Sindi Shkodrani</creatorcontrib><creatorcontrib>Wang, Yu</creatorcontrib><creatorcontrib>Manfredi, Marco</creatorcontrib><creatorcontrib>Baka, Nóra</creatorcontrib><title>United We Learn Better: Harvesting Learning Improvements From Class Hierarchies Across Tasks</title><title>arXiv.org</title><description>Attempts of learning from hierarchical taxonomies in computer vision have been mostly focusing on image classification. Though ways of best harvesting learning improvements from hierarchies in classification are far from being solved, there is a need to target these problems in other vision tasks such as object detection. As progress on the classification side is often dependent on hierarchical cross-entropy losses, novel detection architectures using sigmoid as an output function instead of softmax cannot easily apply these advances, requiring novel methods in detection. In this work we establish a theoretical framework based on probability and set theory for extracting parent predictions and a hierarchical loss that can be used across tasks, showing results across classification and detection benchmarks and opening up the possibility of hierarchical learning for sigmoid-based detection architectures.</description><subject>Classification</subject><subject>Computer vision</subject><subject>Entropy (Information theory)</subject><subject>Hierarchies</subject><subject>Image classification</subject><subject>Learning</subject><subject>Object recognition</subject><subject>Set theory</subject><subject>Taxonomy</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><recordid>eNqNjs0KwjAQhIMgWLTvEPBcSNMfxZsWSwWPihehhLpqa5vobtrnN6IP4GmG-YZhRsyTURQGy1jKCfOJGiGETBcySSKPnY-6tnDhJ-B7UKj5BqwFXPFC4QBka337go_ZdU80A3SgLfEcTcezVhHxogZUWN1rIL6u0LjooOhBMza-qpbA_-mUzfPtISsCN_Pq3XrZmB61Q6V7kyZxKNI0-q_1Bnh3Qv4</recordid><startdate>20210728</startdate><enddate>20210728</enddate><creator>Sindi Shkodrani</creator><creator>Wang, Yu</creator><creator>Manfredi, Marco</creator><creator>Baka, Nóra</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20210728</creationdate><title>United We Learn Better: Harvesting Learning Improvements From Class Hierarchies Across Tasks</title><author>Sindi Shkodrani ; Wang, Yu ; Manfredi, Marco ; Baka, Nóra</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_25565410663</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>Classification</topic><topic>Computer vision</topic><topic>Entropy (Information theory)</topic><topic>Hierarchies</topic><topic>Image classification</topic><topic>Learning</topic><topic>Object recognition</topic><topic>Set theory</topic><topic>Taxonomy</topic><toplevel>online_resources</toplevel><creatorcontrib>Sindi Shkodrani</creatorcontrib><creatorcontrib>Wang, Yu</creatorcontrib><creatorcontrib>Manfredi, Marco</creatorcontrib><creatorcontrib>Baka, Nóra</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni)</collection><collection>ProQuest Central</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central</collection><collection>SciTech Premium Collection (Proquest) (PQ_SDU_P3)</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Sindi Shkodrani</au><au>Wang, Yu</au><au>Manfredi, Marco</au><au>Baka, Nóra</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>United We Learn Better: Harvesting Learning Improvements From Class Hierarchies Across Tasks</atitle><jtitle>arXiv.org</jtitle><date>2021-07-28</date><risdate>2021</risdate><eissn>2331-8422</eissn><abstract>Attempts of learning from hierarchical taxonomies in computer vision have been mostly focusing on image classification. Though ways of best harvesting learning improvements from hierarchies in classification are far from being solved, there is a need to target these problems in other vision tasks such as object detection. As progress on the classification side is often dependent on hierarchical cross-entropy losses, novel detection architectures using sigmoid as an output function instead of softmax cannot easily apply these advances, requiring novel methods in detection. In this work we establish a theoretical framework based on probability and set theory for extracting parent predictions and a hierarchical loss that can be used across tasks, showing results across classification and detection benchmarks and opening up the possibility of hierarchical learning for sigmoid-based detection architectures.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2021-07
issn 2331-8422
language eng
recordid cdi_proquest_journals_2556541066
source Free E- Journals
subjects Classification
Computer vision
Entropy (Information theory)
Hierarchies
Image classification
Learning
Object recognition
Set theory
Taxonomy
title United We Learn Better: Harvesting Learning Improvements From Class Hierarchies Across Tasks
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-07T16%3A53%3A01IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=United%20We%20Learn%20Better:%20Harvesting%20Learning%20Improvements%20From%20Class%20Hierarchies%20Across%20Tasks&rft.jtitle=arXiv.org&rft.au=Sindi%20Shkodrani&rft.date=2021-07-28&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2556541066%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2556541066&rft_id=info:pmid/&rfr_iscdi=true