About Arto Maranjyan Arto Maranjyan Ph.D. Student, Computer Science Events Presented Events Nov 30 - Dec 6, 2025 First Provably Optimal Asynchronous SGD for Homogeneous and Heterogeneous Data Arto Maranjyan, Ph.D. Student, Computer Science Dec 4, 08:30 - 11:00 B4/5 L0 A0215 machine learning optimization asynchronous algorithms Training This thesis introduces a novel framework for asynchronous first-order stochastic optimization centered on heterogeneous worker speeds that remain fast, stable, and even provably optimal. Nov 9 - Nov 15, 2025 First Provably Optimal Asynchronous SGD for Homogeneous and Heterogeneous Data Arto Maranjyan, Ph.D. Student, Computer Science Nov 13, 12:00 - 13:00 B9 L2 R2325 machine learning optimization asynchronous algorithms Training This talk will discuss how to design asynchronous optimization methods that remain fast, stable, and even provably optimal. Oct 26 - Nov 1, 2025 Ringleader ASGD: The First Asynchronous SGD with Optimal Time Complexity under Data Heterogeneity Arto Maranjyan, Ph.D. Student, Computer Science Oct 28, 14:30 - 15:30 B1 L3 R3119 This talk introduces Ringleader ASGD, the first asynchronous SGD algorithm that attains the theoretical lower bounds for parallel first-order stochastic methods in the smooth nonconvex regime, thereby achieving optimal time complexity under data heterogeneity and without restrictive similarity assumptions. Feb 23 - Mar 1, 2025 Ringmaster ASGD: The First Asynchronous SGD with Optimal Time Complexity Arto Maranjyan, Ph.D. Student, Computer Science Feb 27, 12:00 - 13:00 B9, L2, R2325 Asynchronous Stochastic Gradient Descent (Asynchronous SGD) is a cornerstone method for parallelizing learning in distributed machine learning. However, its performance suffers under arbitrarily heterogeneous computation times across workers, leading to suboptimal time complexity and inefficiency as the number of workers scales.
First Provably Optimal Asynchronous SGD for Homogeneous and Heterogeneous Data Arto Maranjyan, Ph.D. Student, Computer Science Dec 4, 08:30 - 11:00 B4/5 L0 A0215 machine learning optimization asynchronous algorithms Training This thesis introduces a novel framework for asynchronous first-order stochastic optimization centered on heterogeneous worker speeds that remain fast, stable, and even provably optimal.
First Provably Optimal Asynchronous SGD for Homogeneous and Heterogeneous Data Arto Maranjyan, Ph.D. Student, Computer Science Nov 13, 12:00 - 13:00 B9 L2 R2325 machine learning optimization asynchronous algorithms Training This talk will discuss how to design asynchronous optimization methods that remain fast, stable, and even provably optimal.
Ringleader ASGD: The First Asynchronous SGD with Optimal Time Complexity under Data Heterogeneity Arto Maranjyan, Ph.D. Student, Computer Science Oct 28, 14:30 - 15:30 B1 L3 R3119 This talk introduces Ringleader ASGD, the first asynchronous SGD algorithm that attains the theoretical lower bounds for parallel first-order stochastic methods in the smooth nonconvex regime, thereby achieving optimal time complexity under data heterogeneity and without restrictive similarity assumptions.
Ringmaster ASGD: The First Asynchronous SGD with Optimal Time Complexity Arto Maranjyan, Ph.D. Student, Computer Science Feb 27, 12:00 - 13:00 B9, L2, R2325 Asynchronous Stochastic Gradient Descent (Asynchronous SGD) is a cornerstone method for parallelizing learning in distributed machine learning. However, its performance suffers under arbitrarily heterogeneous computation times across workers, leading to suboptimal time complexity and inefficiency as the number of workers scales.
Engage LinkedIn GitHub X ShareClipboard Related Sites Computer Science (CS) Optimization and Machine Learning (OML) Related Content Events 4 Related Links Arto Maranjyan personla website Arto Maranjyan profile on Google Scholar
