Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Sparse SVM with Hard-Margin Loss: a Newton-Augmented Lagrangian Method in Reduced Dimensions
Authors: Penghe Zhang, Naihua Xiu, Hou-Duo Qi
JMLR 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive numerical results on both simulated and real data sets demonstrate that the proposed method is fast, produces sparse solution of high accuracy, and can lead to effective reduction on active samples and features when compared with several leading solvers. |
| Researcher Affiliation | Academia | Penghe Zhang EMAIL Department of Data Science and Artificial Intelligence The Hong Kong Polytechnic University Hung Hom, Hong Kong; Naihua Xiu EMAIL School of Mathematics and Statistics Beijing Jiaotong University Beijing, China; Hou-Duo Qi EMAIL Department of Applied Mathematics Department of Data Science and Artificial Intelligence The Hong Kong Polytechnic University Hung Hom, Hong Kong |
| Pseudocode | Yes | Algorithm 1 (i PAL: inexact Proximal Augmented Lagrangian Method); Algorithm 2 (PGN: Projected Gradient-Newton Method) |
| Open Source Code | No | The paper does not explicitly provide a link to its own source code repository or an affirmative statement about releasing the code for the methodology described. |
| Open Datasets | Yes | ID data set Source number of features number of instances all ALLAML feature selection database1 7129 72 ... 1https://jundongl.github.io/scikit-feature/ 2https://archive-beta.ics.uci.edu/datasets 3https://www.openml.org/ 4https://www.refine.bio/ |
| Dataset Splits | Yes | Half of the samples will be chosen as training set, and the rest of the samples are used for testing. We conduct five-fold cross-validation on all the data sets in Tables 2 and 3. |
| Hardware Specification | Yes | extensive numerical experiments will be conducted by using Matlab 2022a on a laptop with 32GB memory and Intel CORE i7 2.6 GHz CPU. |
| Software Dependencies | Yes | extensive numerical experiments will be conducted by using Matlab 2022a |
| Experiment Setup | Yes | We set c1 = c2 = 0.1, γ = 0.1 min{ ai |i [m]}, ϵk = λ/k (41) and η is taken as (24). We adopt (w0, ξ0, z0) = 0 as initial point and i PAL will stop if the following criterion holds wk wk 1 + ξk ξk 1 + zk zk 1 / wk + ξk + zk < 10 3. For i PAL, we set λ = 1, ρ = 1, µ = 10 2. As s will influence the Time and nnz of i PAL, we will set s = 10, 20, 30, 40. |