Time-Smoothed Gradients for Online Forecasting

Time-Smoothed Gradients for Online Forecasting

Here, we study different update rules in stochastic gradient descent (SGD) for online forecasting problems. The selection of the learning rate parameter is critical in SGD. However, it may not be feasible to tune this parameter in online learning. Therefore, it is necessary to have an update rule th...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhu, Tianhao, Aydore, Sergul
Format: Artikel
Sprache:eng
Schlagworte:
Computer Science - Learning
Statistics - Machine Learning
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Zhu, Tianhao
Aydore, Sergul
description Here, we study different update rules in stochastic gradient descent (SGD) for online forecasting problems. The selection of the learning rate parameter is critical in SGD. However, it may not be feasible to tune this parameter in online learning. Therefore, it is necessary to have an update rule that is not sensitive to the selection of the learning parameter. Inspired by the local regret metric that we introduced previously, we propose to use time-smoothed gradients within SGD update. Using the public data set-- GEFCom2014, we validate that our approach yields more stable results than the other existing approaches. Furthermore, we show that such a simple approach is computationally efficient compared to the alternatives.
doi_str_mv 10.48550/arxiv.1905.08850
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_1905_08850</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>1905_08850</sourcerecordid><originalsourceid>FETCH-LOGICAL-a670-bb1f6460d24f09d335bf6ffe671559e24d741395be2379f70c1b698c99cf101a3</originalsourceid><addsrcrecordid>eNotzr1uwjAUQGEvDIjyAEzkBRKu4794rBDQSkgMZI_s-F6wRBLkRFV5ewTtdLajj7EVh0JWSsHGpd_4U3ALqoCqUjBnRR07zM_dMExXDNkhuRCxn8aMhpSd-lvsMdsPCVs3TrG_fLAZuduIy_8uWL3f1duv_Hg6fG8_j7nTBnLvOWmpIZSSwAYhlCdNhNpwpSyWMhjJhVUeS2EsGWi517ZqrW2JA3diwdZ_2ze4uafYufRoXvDmDRdPdE08rg</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Time-Smoothed Gradients for Online Forecasting</title><source>arXiv.org</source><creator>Zhu, Tianhao ; Aydore, Sergul</creator><creatorcontrib>Zhu, Tianhao ; Aydore, Sergul</creatorcontrib><description>Here, we study different update rules in stochastic gradient descent (SGD) for online forecasting problems. The selection of the learning rate parameter is critical in SGD. However, it may not be feasible to tune this parameter in online learning. Therefore, it is necessary to have an update rule that is not sensitive to the selection of the learning parameter. Inspired by the local regret metric that we introduced previously, we propose to use time-smoothed gradients within SGD update. Using the public data set-- GEFCom2014, we validate that our approach yields more stable results than the other existing approaches. Furthermore, we show that such a simple approach is computationally efficient compared to the alternatives.</description><identifier>DOI: 10.48550/arxiv.1905.08850</identifier><language>eng</language><subject>Computer Science - Learning ; Statistics - Machine Learning</subject><creationdate>2019-05</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/1905.08850$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.1905.08850$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Zhu, Tianhao</creatorcontrib><creatorcontrib>Aydore, Sergul</creatorcontrib><title>Time-Smoothed Gradients for Online Forecasting</title><description>Here, we study different update rules in stochastic gradient descent (SGD) for online forecasting problems. The selection of the learning rate parameter is critical in SGD. However, it may not be feasible to tune this parameter in online learning. Therefore, it is necessary to have an update rule that is not sensitive to the selection of the learning parameter. Inspired by the local regret metric that we introduced previously, we propose to use time-smoothed gradients within SGD update. Using the public data set-- GEFCom2014, we validate that our approach yields more stable results than the other existing approaches. Furthermore, we show that such a simple approach is computationally efficient compared to the alternatives.</description><subject>Computer Science - Learning</subject><subject>Statistics - Machine Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotzr1uwjAUQGEvDIjyAEzkBRKu4794rBDQSkgMZI_s-F6wRBLkRFV5ewTtdLajj7EVh0JWSsHGpd_4U3ALqoCqUjBnRR07zM_dMExXDNkhuRCxn8aMhpSd-lvsMdsPCVs3TrG_fLAZuduIy_8uWL3f1duv_Hg6fG8_j7nTBnLvOWmpIZSSwAYhlCdNhNpwpSyWMhjJhVUeS2EsGWi517ZqrW2JA3diwdZ_2ze4uafYufRoXvDmDRdPdE08rg</recordid><startdate>20190521</startdate><enddate>20190521</enddate><creator>Zhu, Tianhao</creator><creator>Aydore, Sergul</creator><scope>AKY</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20190521</creationdate><title>Time-Smoothed Gradients for Online Forecasting</title><author>Zhu, Tianhao ; Aydore, Sergul</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a670-bb1f6460d24f09d335bf6ffe671559e24d741395be2379f70c1b698c99cf101a3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Computer Science - Learning</topic><topic>Statistics - Machine Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Zhu, Tianhao</creatorcontrib><creatorcontrib>Aydore, Sergul</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv Statistics</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Zhu, Tianhao</au><au>Aydore, Sergul</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Time-Smoothed Gradients for Online Forecasting</atitle><date>2019-05-21</date><risdate>2019</risdate><abstract>Here, we study different update rules in stochastic gradient descent (SGD) for online forecasting problems. The selection of the learning rate parameter is critical in SGD. However, it may not be feasible to tune this parameter in online learning. Therefore, it is necessary to have an update rule that is not sensitive to the selection of the learning parameter. Inspired by the local regret metric that we introduced previously, we propose to use time-smoothed gradients within SGD update. Using the public data set-- GEFCom2014, we validate that our approach yields more stable results than the other existing approaches. Furthermore, we show that such a simple approach is computationally efficient compared to the alternatives.</abstract><doi>10.48550/arxiv.1905.08850</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.1905.08850
ispartof
issn
language eng
recordid cdi_arxiv_primary_1905_08850
source arXiv.org
subjects Computer Science - Learning
Statistics - Machine Learning
title Time-Smoothed Gradients for Online Forecasting
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-09T20%3A12%3A02IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Time-Smoothed%20Gradients%20for%20Online%20Forecasting&rft.au=Zhu,%20Tianhao&rft.date=2019-05-21&rft_id=info:doi/10.48550/arxiv.1905.08850&rft_dat=%3Carxiv_GOX%3E1905_08850%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true