Curvature-Exploiting Acceleration of Elastic Net Computations

Authors: Vien Mai, Mikael Johansson

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

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we perform numerical experiments to verify the efficacy of the proposed method on real world data sets (Chang & Lin, 2011; Guyon et al., 2008).
Researcher Affiliation Academia 1Department of Department of Automatic Control, School of Electrical Engineering and Computer Science, Royal Institute of Technology (KTH), Stockholm, Sweden.
Pseudocode Yes Algorithm 1 Randomized Block Lanczos Method (...) Algorithm 2 Inexact Accelerated Scaled Proximal SVRG
Open Source Code No The paper does not provide concrete access to source code (no specific repository link, explicit code release statement, or code in supplementary materials) for the methodology described.
Open Datasets Yes In this section, we perform numerical experiments to verify the efficacy of the proposed method on real world data sets (Chang & Lin, 2011; Guyon et al., 2008).
Dataset Splits No The paper does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) needed to reproduce the data partitioning.
Hardware Specification No The paper does not provide specific hardware details (exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments.
Software Dependencies No The paper mentions software like FISTA, Prox-SVRG, Katyusha1, and BCD, but does not provide specific version numbers for these or any other ancillary software components.
Experiment Setup Yes For each algorithm above, we tune only the step size, from the set η {10k, 2 10k, 5 10k|k {0, 1, 2}}, where η is the theoretical step size, and report the one having smallest objective value. Other hyper-parameters are set to their theory-predicted values. All methods are initialized at 0. For the subproblems in Algorithm 2, we just simply run FISTA with κsub log κsub iterations as discussed previously, without any further tunning steps. The value of r is chosen as a small fraction of d so that the preprocessing time of Algorithm 1 is negligible.