Skip to content

Latest commit

 

History

History
29 lines (21 loc) · 825 Bytes

README.md

File metadata and controls

29 lines (21 loc) · 825 Bytes

Deep Clustering with Convolutional Autoencoders (DCEC)

Keras implementation for ICONIP-2017 paper:

  • Xifeng Guo, Xinwang Liu, En Zhu, Jianping Yin. Deep Clustering with Convolutional Autoencoders. ICONIP 2017.

Usage

  1. Install Keras >=v2.0, scikit-learn and git
    sudo pip install keras scikit-learn
    sudo apt-get install git

  2. Clone the code to local.
    git clone https://github.com/XifengGuo/DCEC.git DCEC

  3. Prepare datasets.

     cd DCEC/data/usps   
     bash ./download_usps.sh   
     cd ../..
    
  4. Run experiment on MNIST.
    python DCEC.py mnist
    The DCEC model will be saved to "results/temp/dcec_model_final.h5".

  5. Run experiment on USPS.
    python DCEC.py usps

  6. Run experiment on MNIST-TEST.
    python DCEC.py mnist-test