Skip to content

Merged Geometrical Homogeneous Clustering for Image Data Reduction. An algorithm to reduce large image datasets maintaining similar accuracy.

Notifications You must be signed in to change notification settings

SoniSiddharth/Merged-GHCIDR

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Merged-GHCIDR 🌟

This code is for the SubsetML ICML 2021 Workshop paper - "Geometrical Homogeneous Clustering for Image Data Reduction"

Requirements

Please download the required modules from the requirements.txt

pip install -r requirements.txt

To create the reduced data

For creating reduced dataset change <dataset> to MNIST/FMNIST/CIFAR10.

For RHC(baseline)

python main.py -variantName RHC -datasetName <dataset>

For RHCKON

python main.py -variantName RHCKON -datasetName <dataset> -KFarthest <K>

For KONCW

python main.py -variantName KONCW -datasetName <dataset> -alpha <alpha>

For CWKC

python main.py -variantName CWKC -datasetName <dataset> -alpha <alpha>

For GHCIDR

python main.py -variantName GHCIDR -datasetName <dataset> -alpha <alpha>

The reduced dataset will be saved in "./datasetPickle" with the name <datasetName>_<variantName>.pickle

To test the reduced data

For testing the reduced dataset <dataset> with variant <variant> change <modelname> to vgg1/fcn.

python vgg1.py -datasetName <dataset> -variantName <variant> -epochs 100 -lr 0.01 -batchSize 64 -fullDataset No

Team Members 🧍

The contributors of this project -

Devvrat Joshi
Janvi Thakkar
Shril Mody
Siddharth Soni