https://github.com/shaharazulay/adversarial-autoencoder-classifier

PyTorch implementations of Non-parametric Unsupervised Classification with Adversarial Autoencoders
https://github.com/shaharazulay/adversarial-autoencoder-classifier

autoencoders deep generative-adversarial-network pytorch unsuper

Last synced: about 1 year ago
JSON representation

PyTorch implementations of Non-parametric Unsupervised Classification with Adversarial Autoencoders

• Host: GitHub
• URL: https://github.com/shaharazulay/adversarial-autoencoder-classifier
• Owner: shaharazulay
• Created: 2019-02-09T06:46:34.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2019-04-26T09:34:07.000Z (about 7 years ago)
• Last Synced: 2025-04-03T07:51:18.661Z (about 1 year ago)
• Topics: autoencoders, deep, generative-adversarial-network, pytorch, unsuper
• Language: Python
• Size: 2.12 MB
• Stars: 11
• Watchers: 1
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: docs/README.rst

Awesome Lists containing this project

README

          
Non Parametric Classification with Advesarial AutoEncoders

============

**PyTorch implementation of Non-parametric Unsupervised Classification with Adversarial Autoencoders.**

Shahar Azulay

|Python27|_ |Python35|_ |License|_ |PyTorch|_ 

|Documentation|_

.. image:: _static/logo.png

.. |PyTorch| image:: https://github.com/pytorch/pytorch/blob/master/docs/source/_static/img/pytorch-logo-flame.svg

.. _PyTorch: https://pytorch.org/

.. |License| image:: https://img.shields.io/badge/license-BSD--3--Clause-brightgreen.svg

.. _License: https://github.com/shaharazulay/traceable-dict/blob/master/LICENSE

    

.. |Python27| image:: https://img.shields.io/badge/python-2.7-blue.svg

.. _Python27:

.. |Python35| image:: https://img.shields.io/badge/python-3.5-blue.svg

.. _Python35:

    

.. |Documentation| image:: _static/readthedocs_logo.jpg

.. _Documentation: https://adversarial-autoencoder-classif.readthedocs.io/en/latest/

*[1] A.Makhzani,  J.Shlens, N.Jaitly, I.Goodfellow, B.Frey: Adversarial Autoencoders, 2016, arXiv:1511.05644v2*

**Usage Examples:**

  Install the module

   

        >>> python setup.py install --user

        

  **Initialize the Datasets**

        

        >>> init_datasets --dir-path 

  **Train a new AAE in an Semi-Supervised setting**

  

        >>> train_semi_supervised --dir-path  --n-epochs 35 --z-size 2 --n-classes 10 --batch-size 100

        loading data started...

        dataset size in use: 3000 [labeled trainset]  47000 [un-labeled trainset]  10000 [validation]

        using configuration:

         {'learning_rates': {'auto_encoder_lr': 0.0008, 'generator_lr': 0.002, 'discriminator_lr': 0.0002, 'info_lr': 1e-05,             'mode_lr': 0.0008, 'disentanglement_lr': 0}, 'model': {'hidden_size': 3000, 'encoder_dropout': 0.2}, 'training':               {'use_mutual_info': False, 'use_mode_decoder': False, 'use_disentanglement': True, 'use_adam_optimization': True,            'use_adversarial_categorial_weights': True, 'lambda_z_l2_regularization': 0.15}}

        current epoch:: [ ===================  ] 99.79%

        ...

  

  .. image:: _static/unsupervised_advesarial_learning_curve.png

  

  **Train a new AAE in a Fully Unsupervised setting**

        >>> train_unsupervised --dir-path  --n-epochs 35 --z-size 2 --n-classes 10 --batch-size 100

        loading data started...

        dataset size in use: 3000 [labeled trainset]  47000 [un-labeled trainset]  10000 [validation]

        ...

        

  **Visualize a trained model using pre-defined visualizations**

        >>> generate_model_visualization --dir-path  --model-dir-path { --mode unsupervised --n-classes 10 --z-size 5

        loading data started...

        dataset size in use: 3000 [labeled trainset]  47000 [un-labeled trainset]  10000 [validation]

        Label 1: 40.2%, Best matching label; 20

        Label 2: 41.9%, Best matching label; 14

        Label 3: 33.0%, Best matching label; 4

        Label 4: 41.1%, Best matching label; 2

        Label 5: 53.8%, Best matching label; 11

        Label 6: 44.3%, Best matching label; 26

        Label 7: 48.6%, Best matching label; 6

        Label 8: 47.6%, Best matching label; 0

        Label 9: 40.1%, Best matching label; 22

        ACCURACY: 0.85%

        ...

  .. image:: _static/modes_and_samples_from_each_label.png

  

  Control the model and training hyper-parameters using a YAML configuration file

        >>> train_unsupervised --dir-path  --config-path  --n-epochs 35 --z-size 2 --n-classes 10 --batch-size 100