A unified variance-reduced accelerated gradient method for convex optimization

Authors: Guanghui Lan, Zhize Li, Yi Zhou

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we provide extensive experimental results to demonstrate the advantages of Varag over several state-of-the-art methods for solving some well-known ML models, e.g., logistic regression, Lasso, etc. We defer the proofs of the main results in Appendix A.
Researcher Affiliation Collaboration Guanghui Lan H. Milton Stewart School of Industrial & Systems Engineering Georgia Institute of Technology Atlanta, GA 30332 george.lan@isye.gatech.edu Zhize Li Institute for Interdisciplinary Information Sciences Tsinghua University Beijing 100084, China zz-li14@mails.tsinghua.edu.cn Yi Zhou IBM Almaden Research Center San Jose, CA 95120 yi.zhou@ibm.com
Pseudocode Yes Algorithm 1 The variance-reduced accelerated gradient (Varag ) method. Algorithm 2 Stochastic accelerated variance-reduced stochastic gradient descent (Stochastic Varag )
Open Source Code No The paper does not provide an explicit statement about releasing source code or a link to a code repository for the methodology described.
Open Datasets Yes For all experiments, we use public real datasets downloaded from UCI Machine Learning Repository [10] and uniform sampling strategy to select fi. Diabetes (m = 1151), Breast Cancer Wisconsin (m = 683), Parkinsons Telemonitoring (m = 5875)
Dataset Splits No The paper does not provide specific details on dataset split percentages, counts, or explicit methodology for training, validation, or test sets.
Hardware Specification No The paper does not provide specific hardware details (e.g., CPU/GPU models, memory) used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers (e.g., Python 3.8, PyTorch 1.9, CPLEX 12.4).
Experiment Setup Yes The algorithmic parameters for SVRG++ and Katyushans are set according to [2] and [1], respectively, and those for Varag are set as in Theorem 1. The algorithmic parameters for SVRG++ and Katyushans are set according to [2] and [1], respectively, and those for Varag are set as in Theorem 2. The algorithmic parameters for FGM and Varag are set according to [21] and Theorem 3, respectively.