<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE ArticleSet PUBLIC "-//NLM//DTD PubMed 2.7//EN" "https://dtd.nlm.nih.gov/ncbi/pubmed/in/PubMed.dtd">
<ArticleSet>
<Article>
<Journal>
				<PublisherName>Ferdowsi University of Mashhad</PublisherName>
				<JournalTitle>Iranian Journal of Numerical Analysis and Optimization</JournalTitle>
				<Issn>2423-6977</Issn>
				<Volume>16</Volume>
				<Issue>Issue 4</Issue>
				<PubDate PubStatus="epublish">
					<Year>2026</Year>
					<Month>12</Month>
					<Day>01</Day>
				</PubDate>
			</Journal>
<ArticleTitle>Practical early stopping for adaptive barrier SGD: balancing stochastic speed with validation accuracy</ArticleTitle>
<VernacularTitle></VernacularTitle>
			<FirstPage>1295</FirstPage>
			<LastPage>1318</LastPage>
			<ELocationID EIdType="pii">47946</ELocationID>
			
<ELocationID EIdType="doi">10.22067/ijnao.2026.96710.1788</ELocationID>
			
			<Language>EN</Language>
<AuthorList>
<Author>
					<FirstName>Abdelouakil</FirstName>
					<LastName>Fillali</LastName>
<Affiliation>Faculty of Mathematics and Informatics, Department of Mathematics, University  Mohamed El Bachir El Ibrahimi of Bordj Bou-Arreridj,Algeria.</Affiliation>
<Identifier Source="ORCID">0009-0002-8907-4876</Identifier>

</Author>
<Author>
					<FirstName>Smail</FirstName>
					<LastName>Addoune</LastName>
<Affiliation>Faculty of Mathematics and Informatics, Department of Mathematics, University  Mohamed El Bachir El Ibrahimi of Bordj Bou-Arreridj,Algeria.</Affiliation>

</Author>
<Author>
					<FirstName>Rebiha</FirstName>
					<LastName>Zeghdane</LastName>
<Affiliation>Faculty of Mathematics and Informatics, Department of Mathematics, University  Mohamed El Bachir El Ibrahimi of Bordj Bou-Arreridj,Algeria.</Affiliation>

</Author>
</AuthorList>
				<PublicationType>Journal Article</PublicationType>
			<History>
				<PubDate PubStatus="received">
					<Year>2025</Year>
					<Month>11</Month>
					<Day>26</Day>
				</PubDate>
			</History>
		    <Abstract>Stochastic Gradient Descent (SGD) algorithms based on adaptive barrier functions are highly efficient for large-scale constrained optimization. However, their inherent stochastic nature leads to the critical ”last- iterate problem”, where update noise causes significant oscillations, making reliance on the final solution a high-stakes gamble. To address this instability, we introduce the Early-Stopped Barrier SGD (EB-SGD) algorithm, a practical approach that replaces reliance on the noisy last iterate with a robust ”best-iterate tracking” mechanism. Our method employs periodic validation, where the cheap stochastic process is temporarily halted to compute a high-quality, deterministic ”Validation Score.” This score is defined as a combination of the full-batch objective function and a large penalty for constraint violation. We maintain the iterate that achieves the best score recorded so far and utilize a ”patience”-based stopping criterion to filter out minor stochastic oscillations and halt the algorithm only upon persistent stagnation. This strategy introduces a deliberate computational trade-off: exchanging many inexpensive stochastic steps for fewer, more expensive validation steps. We prove theoretically that the sequence of best iterates returned by EB-SGD converges almost surely to the unique optimal solution. Furthermore, our numerical experiments confirm the practical superiority of this approach, demonstrating significantly reduced variance (higher stability) and finding a final solution that is substantially more accurate than the baseline last-iterate method.</Abstract>
		<ObjectList>
			<Object Type="keyword">
			<Param Name="value">Constrained optimization</Param>
			</Object>
			<Object Type="keyword">
			<Param Name="value">Stochastic Gradient Descent</Param>
			</Object>
			<Object Type="keyword">
			<Param Name="value">Early Stopping</Param>
			</Object>
			<Object Type="keyword">
			<Param Name="value">Best-Iterate Tracking</Param>
			</Object>
		</ObjectList>
<ArchiveCopySource DocType="pdf">https://ijnao.um.ac.ir/article_47946_26099adf9fe6be27c19d1a0fb68d7985.pdf</ArchiveCopySource>
</Article>
</ArticleSet>
