Loading…

AutoDIAL: Automatic Domain Alignment Layers

Classifiers trained on given databases perform poorly when tested on data acquired in different settings. This is explained in domain adaptation through a shift among distributions of the source and target domains. Attempts to align them have traditionally resulted in works reducing the domain shift...

Full description

Saved in:
Bibliographic Details
Main Authors: Carlucci, Fabio Maria, Porzi, Lorenzo, Caputo, Barbara, Ricci, Elisa, Bulo, Samuel Rota
Format: Conference Proceeding
Language:English
Subjects:
Citations: Items that cite this one
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by cdi_FETCH-LOGICAL-c258t-eec12428e13d41df9e13b12f6d0632053cbc6accfa416c957e492b6bf06832323
cites
container_end_page 5085
container_issue
container_start_page 5077
container_title
container_volume
creator Carlucci, Fabio Maria
Porzi, Lorenzo
Caputo, Barbara
Ricci, Elisa
Bulo, Samuel Rota
description Classifiers trained on given databases perform poorly when tested on data acquired in different settings. This is explained in domain adaptation through a shift among distributions of the source and target domains. Attempts to align them have traditionally resulted in works reducing the domain shift by introducing appropriate loss terms, measuring the discrepancies between source and target distributions, in the objective function. Here we take a different route, proposing to align the learned representations by embedding in any given network specific Domain Alignment Layers, designed to match the source and target feature distributions to a reference one. Opposite to previous works which define a priori in which layers adaptation should be performed, our method is able to automatically learn the degree of feature alignment required at different levels of the deep network. Thorough experiments on different public benchmarks, in the unsupervised setting, confirm the power of our approach.
doi_str_mv 10.1109/ICCV.2017.542
format conference_proceeding
fullrecord <record><control><sourceid>ieee_CHZPO</sourceid><recordid>TN_cdi_ieee_primary_8237804</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>8237804</ieee_id><sourcerecordid>8237804</sourcerecordid><originalsourceid>FETCH-LOGICAL-c258t-eec12428e13d41df9e13b12f6d0632053cbc6accfa416c957e492b6bf06832323</originalsourceid><addsrcrecordid>eNotjD1PwzAUAA0SEqV0ZGLJjhLee_5mi1I-IkXqUlgrx7GRURNQEob-e4JAN9xNx9gNQoEI9r6uqreCAHUhBZ2xjdUGJTcKgZM9ZyviBnItQVyyq2n6AOCWjFqxu_J7_tzWZfOQ_Vbv5uSz7eI0ZOUxvQ99GOascacwTtfsIrrjFDb_XrPXp8d99ZI3u-e6KpvckzRzHoJHEmQC8k5gF-0SLVJUHShOILlvvXLeRydQeSt1EJZa1UZQhtPCmt3-fVMI4fA1pt6Np4Mhrg0I_gO5Bz_q</addsrcrecordid><sourcetype>Publisher</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype></control><display><type>conference_proceeding</type><title>AutoDIAL: Automatic Domain Alignment Layers</title><source>IEEE Xplore All Conference Series</source><creator>Carlucci, Fabio Maria ; Porzi, Lorenzo ; Caputo, Barbara ; Ricci, Elisa ; Bulo, Samuel Rota</creator><creatorcontrib>Carlucci, Fabio Maria ; Porzi, Lorenzo ; Caputo, Barbara ; Ricci, Elisa ; Bulo, Samuel Rota</creatorcontrib><description>Classifiers trained on given databases perform poorly when tested on data acquired in different settings. This is explained in domain adaptation through a shift among distributions of the source and target domains. Attempts to align them have traditionally resulted in works reducing the domain shift by introducing appropriate loss terms, measuring the discrepancies between source and target distributions, in the objective function. Here we take a different route, proposing to align the learned representations by embedding in any given network specific Domain Alignment Layers, designed to match the source and target feature distributions to a reference one. Opposite to previous works which define a priori in which layers adaptation should be performed, our method is able to automatically learn the degree of feature alignment required at different levels of the deep network. Thorough experiments on different public benchmarks, in the unsupervised setting, confirm the power of our approach.</description><identifier>EISSN: 2380-7504</identifier><identifier>EISBN: 9781538610329</identifier><identifier>EISBN: 1538610329</identifier><identifier>DOI: 10.1109/ICCV.2017.542</identifier><identifier>CODEN: IEEPAD</identifier><language>eng</language><publisher>IEEE</publisher><subject>Adaptation models ; Benchmark testing ; Machine learning ; Optimization ; Training ; Visualization</subject><ispartof>2017 IEEE International Conference on Computer Vision (ICCV), 2017, p.5077-5085</ispartof><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c258t-eec12428e13d41df9e13b12f6d0632053cbc6accfa416c957e492b6bf06832323</citedby></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/8237804$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>309,310,780,784,789,790,27925,54555,54932</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/8237804$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Carlucci, Fabio Maria</creatorcontrib><creatorcontrib>Porzi, Lorenzo</creatorcontrib><creatorcontrib>Caputo, Barbara</creatorcontrib><creatorcontrib>Ricci, Elisa</creatorcontrib><creatorcontrib>Bulo, Samuel Rota</creatorcontrib><title>AutoDIAL: Automatic Domain Alignment Layers</title><title>2017 IEEE International Conference on Computer Vision (ICCV)</title><addtitle>ICCV</addtitle><description>Classifiers trained on given databases perform poorly when tested on data acquired in different settings. This is explained in domain adaptation through a shift among distributions of the source and target domains. Attempts to align them have traditionally resulted in works reducing the domain shift by introducing appropriate loss terms, measuring the discrepancies between source and target distributions, in the objective function. Here we take a different route, proposing to align the learned representations by embedding in any given network specific Domain Alignment Layers, designed to match the source and target feature distributions to a reference one. Opposite to previous works which define a priori in which layers adaptation should be performed, our method is able to automatically learn the degree of feature alignment required at different levels of the deep network. Thorough experiments on different public benchmarks, in the unsupervised setting, confirm the power of our approach.</description><subject>Adaptation models</subject><subject>Benchmark testing</subject><subject>Machine learning</subject><subject>Optimization</subject><subject>Training</subject><subject>Visualization</subject><issn>2380-7504</issn><isbn>9781538610329</isbn><isbn>1538610329</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>2017</creationdate><recordtype>conference_proceeding</recordtype><sourceid>6IE</sourceid><recordid>eNotjD1PwzAUAA0SEqV0ZGLJjhLee_5mi1I-IkXqUlgrx7GRURNQEob-e4JAN9xNx9gNQoEI9r6uqreCAHUhBZ2xjdUGJTcKgZM9ZyviBnItQVyyq2n6AOCWjFqxu_J7_tzWZfOQ_Vbv5uSz7eI0ZOUxvQ99GOascacwTtfsIrrjFDb_XrPXp8d99ZI3u-e6KpvckzRzHoJHEmQC8k5gF-0SLVJUHShOILlvvXLeRydQeSt1EJZa1UZQhtPCmt3-fVMI4fA1pt6Np4Mhrg0I_gO5Bz_q</recordid><startdate>201710</startdate><enddate>201710</enddate><creator>Carlucci, Fabio Maria</creator><creator>Porzi, Lorenzo</creator><creator>Caputo, Barbara</creator><creator>Ricci, Elisa</creator><creator>Bulo, Samuel Rota</creator><general>IEEE</general><scope>6IE</scope><scope>6IH</scope><scope>CBEJK</scope><scope>RIE</scope><scope>RIO</scope></search><sort><creationdate>201710</creationdate><title>AutoDIAL: Automatic Domain Alignment Layers</title><author>Carlucci, Fabio Maria ; Porzi, Lorenzo ; Caputo, Barbara ; Ricci, Elisa ; Bulo, Samuel Rota</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c258t-eec12428e13d41df9e13b12f6d0632053cbc6accfa416c957e492b6bf06832323</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>2017</creationdate><topic>Adaptation models</topic><topic>Benchmark testing</topic><topic>Machine learning</topic><topic>Optimization</topic><topic>Training</topic><topic>Visualization</topic><toplevel>online_resources</toplevel><creatorcontrib>Carlucci, Fabio Maria</creatorcontrib><creatorcontrib>Porzi, Lorenzo</creatorcontrib><creatorcontrib>Caputo, Barbara</creatorcontrib><creatorcontrib>Ricci, Elisa</creatorcontrib><creatorcontrib>Bulo, Samuel Rota</creatorcontrib><collection>IEEE Electronic Library (IEL) Conference Proceedings</collection><collection>IEEE Proceedings Order Plan (POP) 1998-present by volume</collection><collection>IEEE Xplore All Conference Proceedings</collection><collection>IEEE Electronic Library Online</collection><collection>IEEE Proceedings Order Plans (POP) 1998-present</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Carlucci, Fabio Maria</au><au>Porzi, Lorenzo</au><au>Caputo, Barbara</au><au>Ricci, Elisa</au><au>Bulo, Samuel Rota</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>AutoDIAL: Automatic Domain Alignment Layers</atitle><btitle>2017 IEEE International Conference on Computer Vision (ICCV)</btitle><stitle>ICCV</stitle><date>2017-10</date><risdate>2017</risdate><spage>5077</spage><epage>5085</epage><pages>5077-5085</pages><eissn>2380-7504</eissn><eisbn>9781538610329</eisbn><eisbn>1538610329</eisbn><coden>IEEPAD</coden><abstract>Classifiers trained on given databases perform poorly when tested on data acquired in different settings. This is explained in domain adaptation through a shift among distributions of the source and target domains. Attempts to align them have traditionally resulted in works reducing the domain shift by introducing appropriate loss terms, measuring the discrepancies between source and target distributions, in the objective function. Here we take a different route, proposing to align the learned representations by embedding in any given network specific Domain Alignment Layers, designed to match the source and target feature distributions to a reference one. Opposite to previous works which define a priori in which layers adaptation should be performed, our method is able to automatically learn the degree of feature alignment required at different levels of the deep network. Thorough experiments on different public benchmarks, in the unsupervised setting, confirm the power of our approach.</abstract><pub>IEEE</pub><doi>10.1109/ICCV.2017.542</doi><tpages>9</tpages><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier EISSN: 2380-7504
ispartof 2017 IEEE International Conference on Computer Vision (ICCV), 2017, p.5077-5085
issn 2380-7504
language eng
recordid cdi_ieee_primary_8237804
source IEEE Xplore All Conference Series
subjects Adaptation models
Benchmark testing
Machine learning
Optimization
Training
Visualization
title AutoDIAL: Automatic Domain Alignment Layers
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T17%3A40%3A23IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-ieee_CHZPO&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=proceeding&rft.atitle=AutoDIAL:%20Automatic%20Domain%20Alignment%20Layers&rft.btitle=2017%20IEEE%20International%20Conference%20on%20Computer%20Vision%20(ICCV)&rft.au=Carlucci,%20Fabio%20Maria&rft.date=2017-10&rft.spage=5077&rft.epage=5085&rft.pages=5077-5085&rft.eissn=2380-7504&rft.coden=IEEPAD&rft_id=info:doi/10.1109/ICCV.2017.542&rft.eisbn=9781538610329&rft.eisbn_list=1538610329&rft_dat=%3Cieee_CHZPO%3E8237804%3C/ieee_CHZPO%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c258t-eec12428e13d41df9e13b12f6d0632053cbc6accfa416c957e492b6bf06832323%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_id=info:pmid/&rft_ieee_id=8237804&rfr_iscdi=true