GV

GeezerVison

The World Wide Web, restored to 800 pixels of glory.

Home|About|Gaming cabinet|Archives|How it works|Starter sites

On this site

  • Sign in
  • ActionsCI/CD and automation
  • CodespacesCloud development environments
  • All Features
  • Documentation
  • Training
  • Certifications
  • Build with Copilot
  • Customize Copilot
  • Integrations
  • Trending repos
  • GitHub Sponsors
  • Developer articles
  • Enterprises

This view

  • Original page
  • New address

GeezerVison is showing github.com โ€” GitHub - MRColorR/SVM_ADMM-Distributed: Project to Solve a SVM fitting problem in MATLAB using a consensus distributed ADMM approach and CVX. Two SMW has been trained and tested on different datasets.

BranchesTags
Open more actions menu

Latest commit

 

History

24 Commits
24 Commits

Folders and files

NameName
Last commit message
Last commit date
Datasets
Datasets
 
 
cvx-w64/cvx
cvx-w64/cvx
 
 
README.md
README.md
 
 
Tesina_RM_Distributed_SVM.pdf
Tesina_RM_Distributed_SVM.pdf
 
 
cvx-w64.zip
cvx-w64.zip
 
 
dataset.xls
dataset.xls
 
 
newData.m
newData.m
 
 
runMe.m
runMe.m
 
 
svm_admm.m
svm_admm.m
 
 

Repository files navigation

SVM_ADMM-Distributed

Compare a distributed support vector machine with a conventional MATLAB implementation.

๐Ÿš€ Run the example ยท ๐Ÿ“š Documentation ยท ๐Ÿซถ Support

Leave a star โญ if you like this project ๐Ÿ™‚ Thank you!

Overview

Solve a SVM fitting problem in MATLAB using a consensus distributed ADMM approach and CVX.

The main focus of this project is to train a linear SVM and show data separated by the best hyperplane in a figure. Then use some of the data not used in the training to show in another figure the performance of the SVM trained in prediction against the real labels. The important thing is that the first SVM is written from scratch resolving the SVM fitting problem starting from the theory so it doesn't use fitcsvm but it solve a distributed convex optimization problem using distributed ADMM with split by data approach and CVX (to research the best parameters for the SVM in order to train it).

Getting started

Start with runMe.m, then use the project report under Documentation for the model and optimization details.

Documentation

  • Tesina_RM_Distributed_SVM.pdf
  • svm_admm.m
  • newData.m
  • cvx-w64/cvx/LICENSE.txt

What's inside

  • Two linear SVMs one distributed, one centralized.
  • Small sample dataset
  • Random dataset generator
  • performance and data plots
  • project document with details on the formulas and theory used (ITA)
  • CVX installer (sept-2022)

The first SVM is decentralized and is written from scratch starting from the fitting problem of SVM solved in a distributed form through ADMM and CVX. The second one is the classic svm made with fitcsvm for comparison.
The code is written so that they can be easily re-trained on different datasets even with different number of features.

Contributing

Ideas, bug reports, documentation fixes and focused changes are welcome through this repository's collaboration channels.

Please include enough context to reproduce a problem or review a change.

Support

๐Ÿซถ Support the project

Your contributions are vital in helping to sustain the development of open-source projects and tools made freely available to everyone. If you find value in my work and wish to show your support, kindly consider making a donation.

โ˜• Support me on Ko-fi ยท ๐Ÿ’› Donate with cryptocurrency

Thank you for helping me keep building and sharing! ๐Ÿ™ See the supporter directory.

License

A project license has not been declared in this repository. Contact the maintainer about reuse permissions.

About

Project to Solve a SVM fitting problem in MATLAB using a consensus distributed ADMM approach and CVX. Two SMW has been trained and tested on different datasets.

Topics

admmadmm-algorithmbigdatacvxdistributed-computingmatlaboptimizationoptimization-algorithmsoptimization-problemprojectsvmsvm-classifiersvm-training

Resources

Readme
Activity

Stars

9 stars

Watchers

1 watching

Forks

0 forks
Report repository

Contributors

Languages

You are visitor 01402503 โ€” GeezerVison does not store the pages you look up.
Best viewed with Netscape Navigator 4.7 or Internet Explorer 6 at 800ร—600. No cookies required.