On the Convergence of Nesterov’s Accelerated Gradient Method in Stochastic Settings

Authors: Mahmoud Assran, Mike Rabbat

ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In Figure 1 we visualize runs of the ASG method on a least-squares regression problem for different problem condition numbers Q. ...The left plots in each sub-figure depict theoretical predictions from Theorem 1, while the right plots in each sub-figure depict empirical results.
Researcher Affiliation Collaboration 1Department of Electrical & Computer Engineering, Mc Gill University, Montreal, QC, Canada 2Facebook AI Research, Montreal, QC, Canada 3Mila Quebec Artificial Intelligence Institute, Montreal, QC, Canada.
Pseudocode No The paper describes the AG method using equations (2) and (3) but does not provide a formally structured pseudocode or algorithm block.
Open Source Code No The paper does not provide any concrete access information (e.g., repository link, explicit statement of code release) for the source code of the methodology described.
Open Datasets No The paper uses 'least-squares regression problem' and describes 'randomly generated least-squares problems' for its numerical experiments, but it does not provide concrete access information (specific link, DOI, repository name, formal citation) for a publicly available or open dataset.
Dataset Splits No The paper conducts numerical experiments on synthetically generated problems or functions (e.g., 'worst-case quadratic function') rather than a dataset with explicitly defined training, validation, and test splits. Therefore, it does not provide specific dataset split information.
Hardware Specification No The paper does not provide specific hardware details (e.g., exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details (e.g., library or solver names with version numbers) needed to replicate the experiment.
Experiment Setup Yes The ASG method with constant step-size and momentum parameters. ... Stochastic gradients are sampled by adding zero-mean Gaussian noise with variance σ2 = 0.0025 to the true gradient. ... The L-smoothness parameter is 100 and the modulus of strong-convexity µ is 0.05. ... with α = 1/L and β = Q 1 Q+1. ... with step-size α = 2/(µ+L).