Quartz: Randomized Dual Coordinate Ascent with Arbitrary Sampling

Authors: Zheng Qu, Peter Richtarik, Tong Zhang

NeurIPS 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section we demonstrate how Quartz specialized to different samplings compares with other methods. All of our experiments are performed with m = 1, for smoothed hinge-loss functions {φi} with γ = 1 and squared L2-regularizer g, see [20]. The experiments were performed on the three datasets reported in Table 3, and three randomly generated large dataset [12] with n = 100, 000 examples, d = 100, 000 features with different sparsity.
Researcher Affiliation Academia Zheng Qu Department of Mathematics The University of Hong Kong Hong Kong zhengqu@maths.hku.hk Peter Richtarik School of Mathematics The University of Edinburgh EH9 3FD, United Kingdom peter.richtarik@ed.ac.uk Tong Zhang Department of Statistics Rutgers University Piscataway, NJ, 08854 tzhang@stat.rutgers.edu
Pseudocode Yes Algorithm 1 Quartz
Open Source Code No The paper does not contain any explicit statement that the source code for the methodology described is publicly available, nor does it provide any links to a code repository.
Open Datasets Yes The experiments were performed on the three datasets reported in Table 3, and three randomly generated large dataset [12] with n = 100, 000 examples, d = 100, 000 features with different sparsity. Table 3 lists 'cov1', 'w8a', 'ijcnn1'.
Dataset Splits No The paper mentions performing experiments on datasets (Table 3) but does not provide specific details on how these datasets were split into training, validation, or test sets (e.g., percentages, sample counts, or explicit standard split references).
Hardware Specification No The paper does not provide specific details about the hardware used to run the experiments, such as GPU models, CPU types, or memory specifications.
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., specific Python, PyTorch, or library versions) that would be needed to reproduce the experiments.
Experiment Setup Yes All of our experiments are performed with m = 1, for smoothed hinge-loss functions {φi} with γ = 1 and squared L2-regularizer g, see [20]. (λ values are also given under each figure: e.g., 'λ = 1e-06')