Stochastic Proximal Gradient Descent with Acceleration Techniques

Authors: Atsushi Nitanda

NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 4 Numerical Experiments In this section, we compare Acc-Prox-SVRG with Prox-SVRG and APG on L1-regularized multiclass logistic regression with the regularization parameter λ. Table 2 provides details of the datasets... Figure 2 compares Acc-Prox-SVRG with Prox-SVRG and APG.
Researcher Affiliation Industry Atsushi Nitanda NTT DATA Mathematical Systems Inc. 1F Shinanomachi Rengakan, 35, Shinanomachi, Shinjuku-ku, Tokyo, 160-0016, Japan nitanda@msi.co.jp
Pseudocode Yes Figure 1: Acc-Prox-SVRG
Open Source Code No The paper does not provide concrete access to source code for the methodology described. No repository link or explicit code release statement is found.
Open Datasets Yes These datasets can be found at the LIBSVM website1. 1http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/
Dataset Splits No Table 2: Details of data sets and regularization parameters. The table provides 'Training size' and 'Testing size' for each dataset (e.g., mnist: Training size 60,000, Testing size 10,000), but does not specify a validation split or methodology for creating such splits.
Hardware Specification No The paper does not provide specific hardware details (exact GPU/CPU models, processor types, or memory amounts) 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 We ran Acc Prox-SVRG using values of η from the range {0.01, 0.05, 0.1, 0.5, 1.0, 5.0, 10.0}, and we chose the best η in each mini-batch setting. For this, we set m = δb (δ {0.1, 1.0, 10}) and βk = b 2 / (b+2) varying b in the set {100, 500, 1000}.