Geometric Descent Method for Convex Composite Minimization
Authors: Shixiang Chen, Shiqian Ma, Wei Liu
NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Numerical results on linear regression and logistic regression with elastic net regularization show that Geo PG compares favorably with Nesterov s accelerated proximal gradient method, especially when the problem is ill-conditioned. |
| Researcher Affiliation | Collaboration | Shixiang Chen1, Shiqian Ma2, and Wei Liu3 1Department of SEEM, The Chinese University of Hong Kong, Hong Kong 2Department of Mathematics, UC Davis, USA 3Tencent AI Lab, China |
| Pseudocode | Yes | Algorithm 1 : The first procedure for finding xk from given x+ k 1 and ck 1. Algorithm 2 : The second procedure for finding xk from given x+ k 1 and ck 1. Algorithm 3 : Geo PG: geometric proximal gradient descent for convex composite minimization. Algorithm 4 : Geo PG with Backtracking (Geo PG-B) Algorithm 5 : L-Geo PG: Limited-memory Geo PG |
| Open Source Code | No | The paper does not provide concrete access to source code for the methodology described. |
| Open Datasets | Yes | We conducted tests on two real datasets downloaded from the LIBSVM repository: a9a, RCV1. We tested Geo PG-B and APG-B for solving (5.2) on the three real datasets a9a, RCV1 and Gisette from LIBSVM |
| Dataset Splits | No | The paper does not provide specific dataset split information (exact percentages, sample counts, or detailed splitting methodology) for training, validation, or test sets. |
| Hardware Specification | Yes | The code was written in Matlab and run on a standard PC with 3.20 GHz I5 Intel microprocessor and 16GB of memory. |
| Software Dependencies | No | The paper only mentions 'Matlab' without a specific version number, and no other software dependencies with versions are listed. |
| Experiment Setup | Yes | The initial points were set to zero. The parameters used in backtracking were set to η = 0.5 and γ = 0.9. In the experiments, we ran Algorithm 2 until the absolute value of φ is smaller than 10 8. |