Alternate Training of Shared and Task-Specific Parameters for Multi-Task Neural Networks

This paper introduces novel alternate training procedures for hard-parameter sharing Multi-Task Neural Networks (MTNNs). Traditional MTNN training faces challenges in managing conflicting loss gradients, often yielding sub-optimal performance. The proposed alternate training method updates shared an...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2023-12
Hauptverfasser: Bellavia, Stefania, Francesco Della Santa, Papini, Alessandra
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 Bellavia, Stefania
Francesco Della Santa
Papini, Alessandra
description This paper introduces novel alternate training procedures for hard-parameter sharing Multi-Task Neural Networks (MTNNs). Traditional MTNN training faces challenges in managing conflicting loss gradients, often yielding sub-optimal performance. The proposed alternate training method updates shared and task-specific weights alternately, exploiting the multi-head architecture of the model. This approach reduces computational costs, enhances training regularization, and improves generalization. Convergence properties similar to those of the classical stochastic gradient method are established. Empirical experiments demonstrate delayed overfitting, improved prediction, and reduced computational demands. In summary, our alternate training procedures offer a promising advancement for the training of hard-parameter sharing MTNNs.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2907597528</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2907597528</sourcerecordid><originalsourceid>FETCH-proquest_journals_29075975283</originalsourceid><addsrcrecordid>eNqNjcEKgkAUAJcgSMp_eNB5wdZMPUYUXYogD93kobu1urm2b6Xfz6AP6DSXGWbCAhHHK56thZixkKiJokhsUpEkccBuW-Ol69BLKBzqTnd3sAquD3SyBuxqKJBafu1lpZWu4IIOn3JMCJR1cBqM1_yrwFkODs0I_7aupQWbKjQkwx_nbHnYF7sj7519DZJ82dhhHBsqRR6lSZ4mIov_sz5ZPUGU</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2907597528</pqid></control><display><type>article</type><title>Alternate Training of Shared and Task-Specific Parameters for Multi-Task Neural Networks</title><source>Freely Accessible Journals_</source><creator>Bellavia, Stefania ; Francesco Della Santa ; Papini, Alessandra</creator><creatorcontrib>Bellavia, Stefania ; Francesco Della Santa ; Papini, Alessandra</creatorcontrib><description>This paper introduces novel alternate training procedures for hard-parameter sharing Multi-Task Neural Networks (MTNNs). Traditional MTNN training faces challenges in managing conflicting loss gradients, often yielding sub-optimal performance. The proposed alternate training method updates shared and task-specific weights alternately, exploiting the multi-head architecture of the model. This approach reduces computational costs, enhances training regularization, and improves generalization. Convergence properties similar to those of the classical stochastic gradient method are established. Empirical experiments demonstrate delayed overfitting, improved prediction, and reduced computational demands. In summary, our alternate training procedures offer a promising advancement for the training of hard-parameter sharing MTNNs.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Computing costs ; Neural networks ; Parameters ; Regularization ; Training</subject><ispartof>arXiv.org, 2023-12</ispartof><rights>2023. 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>776,780</link.rule.ids></links><search><creatorcontrib>Bellavia, Stefania</creatorcontrib><creatorcontrib>Francesco Della Santa</creatorcontrib><creatorcontrib>Papini, Alessandra</creatorcontrib><title>Alternate Training of Shared and Task-Specific Parameters for Multi-Task Neural Networks</title><title>arXiv.org</title><description>This paper introduces novel alternate training procedures for hard-parameter sharing Multi-Task Neural Networks (MTNNs). Traditional MTNN training faces challenges in managing conflicting loss gradients, often yielding sub-optimal performance. The proposed alternate training method updates shared and task-specific weights alternately, exploiting the multi-head architecture of the model. This approach reduces computational costs, enhances training regularization, and improves generalization. Convergence properties similar to those of the classical stochastic gradient method are established. Empirical experiments demonstrate delayed overfitting, improved prediction, and reduced computational demands. In summary, our alternate training procedures offer a promising advancement for the training of hard-parameter sharing MTNNs.</description><subject>Computing costs</subject><subject>Neural networks</subject><subject>Parameters</subject><subject>Regularization</subject><subject>Training</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>BENPR</sourceid><recordid>eNqNjcEKgkAUAJcgSMp_eNB5wdZMPUYUXYogD93kobu1urm2b6Xfz6AP6DSXGWbCAhHHK56thZixkKiJokhsUpEkccBuW-Ol69BLKBzqTnd3sAquD3SyBuxqKJBafu1lpZWu4IIOn3JMCJR1cBqM1_yrwFkODs0I_7aupQWbKjQkwx_nbHnYF7sj7519DZJ82dhhHBsqRR6lSZ4mIov_sz5ZPUGU</recordid><startdate>20231226</startdate><enddate>20231226</enddate><creator>Bellavia, Stefania</creator><creator>Francesco Della Santa</creator><creator>Papini, Alessandra</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>20231226</creationdate><title>Alternate Training of Shared and Task-Specific Parameters for Multi-Task Neural Networks</title><author>Bellavia, Stefania ; Francesco Della Santa ; Papini, Alessandra</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_29075975283</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computing costs</topic><topic>Neural networks</topic><topic>Parameters</topic><topic>Regularization</topic><topic>Training</topic><toplevel>online_resources</toplevel><creatorcontrib>Bellavia, Stefania</creatorcontrib><creatorcontrib>Francesco Della Santa</creatorcontrib><creatorcontrib>Papini, Alessandra</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 Korea</collection><collection>SciTech Premium Collection</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>Bellavia, Stefania</au><au>Francesco Della Santa</au><au>Papini, Alessandra</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Alternate Training of Shared and Task-Specific Parameters for Multi-Task Neural Networks</atitle><jtitle>arXiv.org</jtitle><date>2023-12-26</date><risdate>2023</risdate><eissn>2331-8422</eissn><abstract>This paper introduces novel alternate training procedures for hard-parameter sharing Multi-Task Neural Networks (MTNNs). Traditional MTNN training faces challenges in managing conflicting loss gradients, often yielding sub-optimal performance. The proposed alternate training method updates shared and task-specific weights alternately, exploiting the multi-head architecture of the model. This approach reduces computational costs, enhances training regularization, and improves generalization. Convergence properties similar to those of the classical stochastic gradient method are established. Empirical experiments demonstrate delayed overfitting, improved prediction, and reduced computational demands. In summary, our alternate training procedures offer a promising advancement for the training of hard-parameter sharing MTNNs.</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, 2023-12
issn 2331-8422
language eng
recordid cdi_proquest_journals_2907597528
source Freely Accessible Journals_
subjects Computing costs
Neural networks
Parameters
Regularization
Training
title Alternate Training of Shared and Task-Specific Parameters for Multi-Task Neural Networks
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-27T20%3A46%3A35IST&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=Alternate%20Training%20of%20Shared%20and%20Task-Specific%20Parameters%20for%20Multi-Task%20Neural%20Networks&rft.jtitle=arXiv.org&rft.au=Bellavia,%20Stefania&rft.date=2023-12-26&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2907597528%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2907597528&rft_id=info:pmid/&rfr_iscdi=true