Multiple Descent in the Multiple Random Feature Model

Recent works have demonstrated a double descent phenomenon in over-parameterized learning. Although this phenomenon has been investigated by recent works, it has not been fully understood in theory. In this paper, we investigate the multiple descent phenomenon in a class of multi-component predictio...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Meng, Xuran, Yao, Jianfeng, Cao, Yuan
Format: Artikel
Sprache:eng
Schlagworte:
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 Meng, Xuran
Yao, Jianfeng
Cao, Yuan
description Recent works have demonstrated a double descent phenomenon in over-parameterized learning. Although this phenomenon has been investigated by recent works, it has not been fully understood in theory. In this paper, we investigate the multiple descent phenomenon in a class of multi-component prediction models. We first consider a ''double random feature model'' (DRFM) concatenating two types of random features, and study the excess risk achieved by the DRFM in ridge regression. We calculate the precise limit of the excess risk under the high dimensional framework where the training sample size, the dimension of data, and the dimension of random features tend to infinity proportionally. Based on the calculation, we further theoretically demonstrate that the risk curves of DRFMs can exhibit triple descent. We then provide a thorough experimental study to verify our theory. At last, we extend our study to the ''multiple random feature model'' (MRFM), and show that MRFMs ensembling $K$ types of random features may exhibit $(K+1)$-fold descent. Our analysis points out that risk curves with a specific number of descent generally exist in learning multi-component prediction models.
doi_str_mv 10.48550/arxiv.2208.09897
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2208_09897</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2208_09897</sourcerecordid><originalsourceid>FETCH-LOGICAL-a677-ad424daf95571ba771557555afb09329d1c66bd8bd559bc44b635ad31de3c44a3</originalsourceid><addsrcrecordid>eNo9jssKwjAURLNxIeoHuDI_0Jo0uU2zFN-gCOK-3PSmWKhVahX9e-sDVzNzFsNhbChFqBMAMcb6UdzDKBJJKGxiTZfB9lY2xaX0fOavma8aXlS8OXr-53us6HziC4_NrW75mXzZZ50cy6sf_LLHDov5YboKNrvlejrZBBgbEyDpSBPmFsBIh8bItgAA5k5YFVmSWRw7ShwBWJdp7WIFSEqSV-1C1WOj7-3HO73UxQnrZ_r2Tz_-6gVxrz97</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Multiple Descent in the Multiple Random Feature Model</title><source>arXiv.org</source><creator>Meng, Xuran ; Yao, Jianfeng ; Cao, Yuan</creator><creatorcontrib>Meng, Xuran ; Yao, Jianfeng ; Cao, Yuan</creatorcontrib><description>Recent works have demonstrated a double descent phenomenon in over-parameterized learning. Although this phenomenon has been investigated by recent works, it has not been fully understood in theory. In this paper, we investigate the multiple descent phenomenon in a class of multi-component prediction models. We first consider a ''double random feature model'' (DRFM) concatenating two types of random features, and study the excess risk achieved by the DRFM in ridge regression. We calculate the precise limit of the excess risk under the high dimensional framework where the training sample size, the dimension of data, and the dimension of random features tend to infinity proportionally. Based on the calculation, we further theoretically demonstrate that the risk curves of DRFMs can exhibit triple descent. We then provide a thorough experimental study to verify our theory. At last, we extend our study to the ''multiple random feature model'' (MRFM), and show that MRFMs ensembling $K$ types of random features may exhibit $(K+1)$-fold descent. Our analysis points out that risk curves with a specific number of descent generally exist in learning multi-component prediction models.</description><identifier>DOI: 10.48550/arxiv.2208.09897</identifier><language>eng</language><subject>Computer Science - Learning ; Mathematics - Statistics Theory ; Statistics - Machine Learning ; Statistics - Theory</subject><creationdate>2022-08</creationdate><rights>http://creativecommons.org/licenses/by-nc-nd/4.0</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/2208.09897$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2208.09897$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Meng, Xuran</creatorcontrib><creatorcontrib>Yao, Jianfeng</creatorcontrib><creatorcontrib>Cao, Yuan</creatorcontrib><title>Multiple Descent in the Multiple Random Feature Model</title><description>Recent works have demonstrated a double descent phenomenon in over-parameterized learning. Although this phenomenon has been investigated by recent works, it has not been fully understood in theory. In this paper, we investigate the multiple descent phenomenon in a class of multi-component prediction models. We first consider a ''double random feature model'' (DRFM) concatenating two types of random features, and study the excess risk achieved by the DRFM in ridge regression. We calculate the precise limit of the excess risk under the high dimensional framework where the training sample size, the dimension of data, and the dimension of random features tend to infinity proportionally. Based on the calculation, we further theoretically demonstrate that the risk curves of DRFMs can exhibit triple descent. We then provide a thorough experimental study to verify our theory. At last, we extend our study to the ''multiple random feature model'' (MRFM), and show that MRFMs ensembling $K$ types of random features may exhibit $(K+1)$-fold descent. Our analysis points out that risk curves with a specific number of descent generally exist in learning multi-component prediction models.</description><subject>Computer Science - Learning</subject><subject>Mathematics - Statistics Theory</subject><subject>Statistics - Machine Learning</subject><subject>Statistics - Theory</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNo9jssKwjAURLNxIeoHuDI_0Jo0uU2zFN-gCOK-3PSmWKhVahX9e-sDVzNzFsNhbChFqBMAMcb6UdzDKBJJKGxiTZfB9lY2xaX0fOavma8aXlS8OXr-53us6HziC4_NrW75mXzZZ50cy6sf_LLHDov5YboKNrvlejrZBBgbEyDpSBPmFsBIh8bItgAA5k5YFVmSWRw7ShwBWJdp7WIFSEqSV-1C1WOj7-3HO73UxQnrZ_r2Tz_-6gVxrz97</recordid><startdate>20220821</startdate><enddate>20220821</enddate><creator>Meng, Xuran</creator><creator>Yao, Jianfeng</creator><creator>Cao, Yuan</creator><scope>AKY</scope><scope>AKZ</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20220821</creationdate><title>Multiple Descent in the Multiple Random Feature Model</title><author>Meng, Xuran ; Yao, Jianfeng ; Cao, Yuan</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a677-ad424daf95571ba771557555afb09329d1c66bd8bd559bc44b635ad31de3c44a3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Computer Science - Learning</topic><topic>Mathematics - Statistics Theory</topic><topic>Statistics - Machine Learning</topic><topic>Statistics - Theory</topic><toplevel>online_resources</toplevel><creatorcontrib>Meng, Xuran</creatorcontrib><creatorcontrib>Yao, Jianfeng</creatorcontrib><creatorcontrib>Cao, Yuan</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv Mathematics</collection><collection>arXiv Statistics</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Meng, Xuran</au><au>Yao, Jianfeng</au><au>Cao, Yuan</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Multiple Descent in the Multiple Random Feature Model</atitle><date>2022-08-21</date><risdate>2022</risdate><abstract>Recent works have demonstrated a double descent phenomenon in over-parameterized learning. Although this phenomenon has been investigated by recent works, it has not been fully understood in theory. In this paper, we investigate the multiple descent phenomenon in a class of multi-component prediction models. We first consider a ''double random feature model'' (DRFM) concatenating two types of random features, and study the excess risk achieved by the DRFM in ridge regression. We calculate the precise limit of the excess risk under the high dimensional framework where the training sample size, the dimension of data, and the dimension of random features tend to infinity proportionally. Based on the calculation, we further theoretically demonstrate that the risk curves of DRFMs can exhibit triple descent. We then provide a thorough experimental study to verify our theory. At last, we extend our study to the ''multiple random feature model'' (MRFM), and show that MRFMs ensembling $K$ types of random features may exhibit $(K+1)$-fold descent. Our analysis points out that risk curves with a specific number of descent generally exist in learning multi-component prediction models.</abstract><doi>10.48550/arxiv.2208.09897</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2208.09897
ispartof
issn
language eng
recordid cdi_arxiv_primary_2208_09897
source arXiv.org
subjects Computer Science - Learning
Mathematics - Statistics Theory
Statistics - Machine Learning
Statistics - Theory
title Multiple Descent in the Multiple Random Feature Model
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-10T17%3A25%3A29IST&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=Multiple%20Descent%20in%20the%20Multiple%20Random%20Feature%20Model&rft.au=Meng,%20Xuran&rft.date=2022-08-21&rft_id=info:doi/10.48550/arxiv.2208.09897&rft_dat=%3Carxiv_GOX%3E2208_09897%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