lizq66 / DEKM-Pytorch

A convenient "DEKM" implemented in PyTorch

Overview

• Algorithm of Deep learning clustering -> Deep Embedded K-Means Clustering
• official code -> https://github.com/spdj2271/DEKM

Result (MNIST)

• Training Result
  Average accuracy of training data can reach over 85%, maximum can reach over 92%.
  model weight is at ./weight

• Test Result

Usage

python train.py -pre_epoch 15 -epoch 10 -k 10

• if your cuda memory is not enough, you should use less training data:
  add the command parameter -take, -take 0.8 will only use 80% training data.

python train.py -pre_epoch 15 -epoch 10 -k 10 -take 0.8

Note

This repository is mainly for academic purposes.