SDNA: Stochastic Dual Newton Ascent for Empirical Risk Minimization

Authors: Zheng Qu, Peter Richtarik, Martin Takac, Olivier Fercoq

ICML 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 7. Numerical Experiments In our first experiment (Figure 1) we compare SDNA and our new minibatch version of SDCA on two real (mushrooms: d = 112, n = 8, 124; cov: d = 54, n = 522, 911) and one synthetic (d = 1, 024, n = 2, 048) dataset.
Researcher Affiliation Academia Zheng Qu ZHENGQU@HKU.HK Department of Mathematics, The University of Hong Kong, Hong Kong Peter Richt arik PETER.RICHTARIK@ED.AC.UK School of Mathematics, The University of Edinburgh, UK Martin Tak aˇc TAKAC.MT@GMAIL.COM Industrial and Systems Engineering, Lehigh University, USA Olivier Fercoq OLIVIER.FERCOQ@TELECOM-PARISTECH.FR LTCI, CNRS, T el ecom Paris-Tech, Universit e Paris-Saclay, France
Pseudocode Yes Algorithm 1 Proximal Overlapping-Block CD
Open Source Code No The paper does not contain any statements about making the source code for their proposed method publicly available, nor does it provide a link to a code repository.
Open Datasets Yes In our first experiment (Figure 1) we compare SDNA and our new minibatch version of SDCA on two real (mushrooms: d = 112, n = 8, 124; cov: d = 54, n = 522, 911) and one synthetic (d = 1, 024, n = 2, 048) dataset.
Dataset Splits No The paper states the use of 'mushrooms', 'cov', and a 'synthetic' dataset, but it does not specify any training, validation, or test dataset splits (e.g., percentages or sample counts).
Hardware Specification No The paper mentions that 'all experiments are done using a single core' but does not provide any specific details about the hardware used, such as GPU models, CPU models, or memory specifications.
Software Dependencies No The paper does not provide any specific software dependency details, such as library names with version numbers (e.g., Python 3.x, TensorFlow x.x, PyTorch x.x).
Experiment Setup Yes In both cases, we used λ = 1/n as the regularization parameter and g(w) = 1/2 ||w||^2_2. Comparison of SDNA and SDCA for minibatch sizes τ = 1, 32, 256. Runtime of SDNA for minibatch sizes τ = 1, 4, 16, 32, 64.