LSEMINK
LSEMINK
This repository contains a Matlab implementation of LSEMINK, a modified Newton-Krylov method introduced in
@article{Kan2023LSEMINK,
title={LSEMINK: A Modified Newton-Krylov Method for Log-Sum-Exp
Minimization},
author={Kan, Kelvin and Nagy, James G and Ruthotto, Lars},
journal={arXiv preprint arXiv:2307.04871},
year={2023},
url = "https://arxiv.org/abs/2307.04871",
}
This repository is also available on Mathworks File Exchange
Getting Started
See minimalExample.m
Reproduce Results in the Paper
To run the experiments, we have five driver codes (four for multinomial logistic regression and one for geometric programming):
1) Main_small_MNIST.m
2) Main_small_CIFAR10.m
3) Main_RFM_MNIST.m
4) Main_AlexNet_CIFAR10.m
5) Main_GeometricProgramming.m
You will also need to install Meganet and CVX.
Dependencies
The LSEMINK method is developed with MATLAB R2022a and should run with newer versions as well. No MATLAB toolboxes are required.
The following packages are required to repeat the image classification experiments:
- MATLAB’s deep neural network toolbox (this is for the transfer learning component in the CIFAR-10 experiments).
- Meganet
To reproduce the geometric programming experiments, you will need CVX.
Acknowledgments
This material is in part based upon work supported by the US National Science Foundation Grants DMS-1751636 and DMS-2038118, the US AFOSR grant FA9550-18-1-0167, and US DOE Office of Advanced Scientific Computing Research Field Work Proposal 20-023231. Any opinions, findings, conclusions, or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the funding agencies.
Cite As
Kai Fung Kan (2024). LSEMINK (https://github.com/KelvinKan/LSEMINK/releases/tag/v1.0.0), GitHub. Retrieved .
MATLAB Release Compatibility
Platform Compatibility
Windows macOS LinuxTags
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!Discover Live Editor
Create scripts with code, output, and formatted text in a single executable document.
examples/GeometricProgramming
examples/ImageClassification
Version | Published | Release Notes | |
---|---|---|---|
1.0.0 |