Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385967628> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W4385967628 abstract "We study the online overlapping batch-means covariance estimator for Stochastic Gradient Descent (SGD) under Markovian sampling. We show that the convergence rates of the covariance estimator are $Obig(sqrt{d},n^{-1/8}(log n)^{1/4}big)$ and $Obig(sqrt{d},n^{-1/8}big)$ under state-dependent and state-independent Markovian sampling, respectively, with $d$ representing dimensionality and $n$ denoting the number of observations or SGD iterations. Remarkably, these rates match the best-known convergence rate previously established for the independent and identically distributed ($iid$) case by cite{zhu2021online}, up to logarithmic factors. Our analysis overcomes significant challenges that arise due to Markovian sampling, leading to the introduction of additional error terms and complex dependencies between the blocks of the batch-means covariance estimator. Moreover, we establish the convergence rate for the first four moments of the $ell_2$ norm of the error of SGD dynamics under state-dependent Markovian data, which holds potential interest as an independent result. To validate our theoretical findings, we provide numerical illustrations to derive confidence intervals for SGD when training linear and logistic regression models under Markovian sampling. Additionally, we apply our approach to tackle the intriguing problem of strategic classification with logistic regression, where adversaries can adaptively modify features during the training process to increase their chances of being classified in a specific target class." @default.
- W4385967628 created "2023-08-19" @default.
- W4385967628 creator A5014500490 @default.
- W4385967628 creator A5036909076 @default.
- W4385967628 date "2023-08-02" @default.
- W4385967628 modified "2023-09-23" @default.
- W4385967628 title "Online covariance estimation for stochastic gradient descent under Markovian sampling" @default.
- W4385967628 doi "https://doi.org/10.48550/arxiv.2308.01481" @default.
- W4385967628 hasPublicationYear "2023" @default.
- W4385967628 type Work @default.
- W4385967628 citedByCount "0" @default.
- W4385967628 crossrefType "posted-content" @default.
- W4385967628 hasAuthorship W4385967628A5014500490 @default.
- W4385967628 hasAuthorship W4385967628A5036909076 @default.
- W4385967628 hasBestOaLocation W43859676281 @default.
- W4385967628 hasConcept C105795698 @default.
- W4385967628 hasConcept C106131492 @default.
- W4385967628 hasConcept C111030470 @default.
- W4385967628 hasConcept C122123141 @default.
- W4385967628 hasConcept C127162648 @default.
- W4385967628 hasConcept C136764020 @default.
- W4385967628 hasConcept C140779682 @default.
- W4385967628 hasConcept C141513077 @default.
- W4385967628 hasConcept C154945302 @default.
- W4385967628 hasConcept C159886148 @default.
- W4385967628 hasConcept C178650346 @default.
- W4385967628 hasConcept C185429906 @default.
- W4385967628 hasConcept C206688291 @default.
- W4385967628 hasConcept C28826006 @default.
- W4385967628 hasConcept C31258907 @default.
- W4385967628 hasConcept C31972630 @default.
- W4385967628 hasConcept C33923547 @default.
- W4385967628 hasConcept C37616216 @default.
- W4385967628 hasConcept C41008148 @default.
- W4385967628 hasConcept C50644808 @default.
- W4385967628 hasConcept C57869625 @default.
- W4385967628 hasConceptScore W4385967628C105795698 @default.
- W4385967628 hasConceptScore W4385967628C106131492 @default.
- W4385967628 hasConceptScore W4385967628C111030470 @default.
- W4385967628 hasConceptScore W4385967628C122123141 @default.
- W4385967628 hasConceptScore W4385967628C127162648 @default.
- W4385967628 hasConceptScore W4385967628C136764020 @default.
- W4385967628 hasConceptScore W4385967628C140779682 @default.
- W4385967628 hasConceptScore W4385967628C141513077 @default.
- W4385967628 hasConceptScore W4385967628C154945302 @default.
- W4385967628 hasConceptScore W4385967628C159886148 @default.
- W4385967628 hasConceptScore W4385967628C178650346 @default.
- W4385967628 hasConceptScore W4385967628C185429906 @default.
- W4385967628 hasConceptScore W4385967628C206688291 @default.
- W4385967628 hasConceptScore W4385967628C28826006 @default.
- W4385967628 hasConceptScore W4385967628C31258907 @default.
- W4385967628 hasConceptScore W4385967628C31972630 @default.
- W4385967628 hasConceptScore W4385967628C33923547 @default.
- W4385967628 hasConceptScore W4385967628C37616216 @default.
- W4385967628 hasConceptScore W4385967628C41008148 @default.
- W4385967628 hasConceptScore W4385967628C50644808 @default.
- W4385967628 hasConceptScore W4385967628C57869625 @default.
- W4385967628 hasLocation W43859676281 @default.
- W4385967628 hasOpenAccess W4385967628 @default.
- W4385967628 hasPrimaryLocation W43859676281 @default.
- W4385967628 hasRelatedWork W1827694088 @default.
- W4385967628 hasRelatedWork W1988224349 @default.
- W4385967628 hasRelatedWork W2017052259 @default.
- W4385967628 hasRelatedWork W2021298725 @default.
- W4385967628 hasRelatedWork W2022232878 @default.
- W4385967628 hasRelatedWork W2266110460 @default.
- W4385967628 hasRelatedWork W4225775431 @default.
- W4385967628 hasRelatedWork W4246387931 @default.
- W4385967628 hasRelatedWork W4280650596 @default.
- W4385967628 hasRelatedWork W4312447901 @default.
- W4385967628 isParatext "false" @default.
- W4385967628 isRetracted "false" @default.
- W4385967628 workType "article" @default.