Statistical and Computational Trade-offs in Variational Inference: A Case Study in Inferential Model Selection

Variational inference has recently emerged as a popular alternative to the classical Markov chain Monte Carlo (MCMC) in large-scale Bayesian inference. The core idea is to trade statistical accuracy for computational efficiency. In this work, we study these statistical and computational trade-offs i...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Bhatia, Kush, Kuang, Nikki Lijing, Ma, Yi-An, Wang, Yixin
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 Bhatia, Kush
Kuang, Nikki Lijing
Ma, Yi-An
Wang, Yixin
description Variational inference has recently emerged as a popular alternative to the classical Markov chain Monte Carlo (MCMC) in large-scale Bayesian inference. The core idea is to trade statistical accuracy for computational efficiency. In this work, we study these statistical and computational trade-offs in variational inference via a case study in inferential model selection. Focusing on Gaussian inferential models (or variational approximating families) with diagonal plus low-rank precision matrices, we initiate a theoretical study of the trade-offs in two aspects, Bayesian posterior inference error and frequentist uncertainty quantification error. From the Bayesian posterior inference perspective, we characterize the error of the variational posterior relative to the exact posterior. We prove that, given a fixed computation budget, a lower-rank inferential model produces variational posteriors with a higher statistical approximation error, but a lower computational error; it reduces variance in stochastic optimization and, in turn, accelerates convergence. From the frequentist uncertainty quantification perspective, we consider the precision matrix of the variational posterior as an uncertainty estimate, which involves an additional statistical error originating from the sampling uncertainty of the data. As a consequence, for small datasets, the inferential model need not be full-rank to achieve optimal estimation error (even with unlimited computation budget).
doi_str_mv 10.48550/arxiv.2207.11208
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2207_11208</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2207_11208</sourcerecordid><originalsourceid>FETCH-LOGICAL-a678-c17d35987b61b54d8cc3e8c119e27ec3839cddfa4e63ee4e99b68297543ad7043</originalsourceid><addsrcrecordid>eNo1j7tqwzAUhrV0KEkfoFP1AnYtS7KkbMH0EkjpYNPVHEtHIHDkIDulefvGaTr98N_gI-SRFbnQUhbPkH7Cd16WhcoZKwt9T2IzwxymOVgYKERH6_FwPC3eGC9Om8BhNno_0RDpF6Twn-yix4TR4oZuaQ0T0mY-ufNSu0VzuNQ-RocDbXBAuwzX5M7DMOHDTVekfX1p6_ds__m2q7f7DCqlM8uU49Jo1Vesl8Jpazlqy5jBUqHlmhvrnAeBFUcUaExf6dIoKTg4VQi-Ik9_t1fg7pjCAdK5W8C7Kzj_BfpuVF4</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Statistical and Computational Trade-offs in Variational Inference: A Case Study in Inferential Model Selection</title><source>arXiv.org</source><creator>Bhatia, Kush ; Kuang, Nikki Lijing ; Ma, Yi-An ; Wang, Yixin</creator><creatorcontrib>Bhatia, Kush ; Kuang, Nikki Lijing ; Ma, Yi-An ; Wang, Yixin</creatorcontrib><description>Variational inference has recently emerged as a popular alternative to the classical Markov chain Monte Carlo (MCMC) in large-scale Bayesian inference. The core idea is to trade statistical accuracy for computational efficiency. In this work, we study these statistical and computational trade-offs in variational inference via a case study in inferential model selection. Focusing on Gaussian inferential models (or variational approximating families) with diagonal plus low-rank precision matrices, we initiate a theoretical study of the trade-offs in two aspects, Bayesian posterior inference error and frequentist uncertainty quantification error. From the Bayesian posterior inference perspective, we characterize the error of the variational posterior relative to the exact posterior. We prove that, given a fixed computation budget, a lower-rank inferential model produces variational posteriors with a higher statistical approximation error, but a lower computational error; it reduces variance in stochastic optimization and, in turn, accelerates convergence. From the frequentist uncertainty quantification perspective, we consider the precision matrix of the variational posterior as an uncertainty estimate, which involves an additional statistical error originating from the sampling uncertainty of the data. As a consequence, for small datasets, the inferential model need not be full-rank to achieve optimal estimation error (even with unlimited computation budget).</description><identifier>DOI: 10.48550/arxiv.2207.11208</identifier><language>eng</language><subject>Computer Science - Learning ; Statistics - Machine Learning</subject><creationdate>2022-07</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>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2207.11208$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2207.11208$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Bhatia, Kush</creatorcontrib><creatorcontrib>Kuang, Nikki Lijing</creatorcontrib><creatorcontrib>Ma, Yi-An</creatorcontrib><creatorcontrib>Wang, Yixin</creatorcontrib><title>Statistical and Computational Trade-offs in Variational Inference: A Case Study in Inferential Model Selection</title><description>Variational inference has recently emerged as a popular alternative to the classical Markov chain Monte Carlo (MCMC) in large-scale Bayesian inference. The core idea is to trade statistical accuracy for computational efficiency. In this work, we study these statistical and computational trade-offs in variational inference via a case study in inferential model selection. Focusing on Gaussian inferential models (or variational approximating families) with diagonal plus low-rank precision matrices, we initiate a theoretical study of the trade-offs in two aspects, Bayesian posterior inference error and frequentist uncertainty quantification error. From the Bayesian posterior inference perspective, we characterize the error of the variational posterior relative to the exact posterior. We prove that, given a fixed computation budget, a lower-rank inferential model produces variational posteriors with a higher statistical approximation error, but a lower computational error; it reduces variance in stochastic optimization and, in turn, accelerates convergence. From the frequentist uncertainty quantification perspective, we consider the precision matrix of the variational posterior as an uncertainty estimate, which involves an additional statistical error originating from the sampling uncertainty of the data. As a consequence, for small datasets, the inferential model need not be full-rank to achieve optimal estimation error (even with unlimited computation budget).</description><subject>Computer Science - Learning</subject><subject>Statistics - Machine Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNo1j7tqwzAUhrV0KEkfoFP1AnYtS7KkbMH0EkjpYNPVHEtHIHDkIDulefvGaTr98N_gI-SRFbnQUhbPkH7Cd16WhcoZKwt9T2IzwxymOVgYKERH6_FwPC3eGC9Om8BhNno_0RDpF6Twn-yix4TR4oZuaQ0T0mY-ufNSu0VzuNQ-RocDbXBAuwzX5M7DMOHDTVekfX1p6_ds__m2q7f7DCqlM8uU49Jo1Vesl8Jpazlqy5jBUqHlmhvrnAeBFUcUaExf6dIoKTg4VQi-Ik9_t1fg7pjCAdK5W8C7Kzj_BfpuVF4</recordid><startdate>20220722</startdate><enddate>20220722</enddate><creator>Bhatia, Kush</creator><creator>Kuang, Nikki Lijing</creator><creator>Ma, Yi-An</creator><creator>Wang, Yixin</creator><scope>AKY</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20220722</creationdate><title>Statistical and Computational Trade-offs in Variational Inference: A Case Study in Inferential Model Selection</title><author>Bhatia, Kush ; Kuang, Nikki Lijing ; Ma, Yi-An ; Wang, Yixin</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a678-c17d35987b61b54d8cc3e8c119e27ec3839cddfa4e63ee4e99b68297543ad7043</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Computer Science - Learning</topic><topic>Statistics - Machine Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Bhatia, Kush</creatorcontrib><creatorcontrib>Kuang, Nikki Lijing</creatorcontrib><creatorcontrib>Ma, Yi-An</creatorcontrib><creatorcontrib>Wang, Yixin</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>Bhatia, Kush</au><au>Kuang, Nikki Lijing</au><au>Ma, Yi-An</au><au>Wang, Yixin</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Statistical and Computational Trade-offs in Variational Inference: A Case Study in Inferential Model Selection</atitle><date>2022-07-22</date><risdate>2022</risdate><abstract>Variational inference has recently emerged as a popular alternative to the classical Markov chain Monte Carlo (MCMC) in large-scale Bayesian inference. The core idea is to trade statistical accuracy for computational efficiency. In this work, we study these statistical and computational trade-offs in variational inference via a case study in inferential model selection. Focusing on Gaussian inferential models (or variational approximating families) with diagonal plus low-rank precision matrices, we initiate a theoretical study of the trade-offs in two aspects, Bayesian posterior inference error and frequentist uncertainty quantification error. From the Bayesian posterior inference perspective, we characterize the error of the variational posterior relative to the exact posterior. We prove that, given a fixed computation budget, a lower-rank inferential model produces variational posteriors with a higher statistical approximation error, but a lower computational error; it reduces variance in stochastic optimization and, in turn, accelerates convergence. From the frequentist uncertainty quantification perspective, we consider the precision matrix of the variational posterior as an uncertainty estimate, which involves an additional statistical error originating from the sampling uncertainty of the data. As a consequence, for small datasets, the inferential model need not be full-rank to achieve optimal estimation error (even with unlimited computation budget).</abstract><doi>10.48550/arxiv.2207.11208</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2207.11208
ispartof
issn
language eng
recordid cdi_arxiv_primary_2207_11208
source arXiv.org
subjects Computer Science - Learning
Statistics - Machine Learning
title Statistical and Computational Trade-offs in Variational Inference: A Case Study in Inferential Model Selection
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-28T12%3A36%3A54IST&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=Statistical%20and%20Computational%20Trade-offs%20in%20Variational%20Inference:%20A%20Case%20Study%20in%20Inferential%20Model%20Selection&rft.au=Bhatia,%20Kush&rft.date=2022-07-22&rft_id=info:doi/10.48550/arxiv.2207.11208&rft_dat=%3Carxiv_GOX%3E2207_11208%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