https://github.com/vislearn/IB-INN

Code for the paper "Training Normalizing Flows with the Information Bottleneck for Competitive Generative Classification" (2020)
https://github.com/vislearn/IB-INN

Last synced: 2 months ago
JSON representation

Code for the paper "Training Normalizing Flows with the Information Bottleneck for Competitive Generative Classification" (2020)

• Host: GitHub
• URL: https://github.com/vislearn/IB-INN
• Owner: vislearn
• Created: 2020-02-19T16:19:09.000Z (almost 5 years ago)
• Default Branch: master
• Last Pushed: 2024-05-10T12:09:35.000Z (9 months ago)
• Last Synced: 2024-05-18T22:49:47.244Z (8 months ago)
• Language: Python
• Homepage:
• Size: 89.8 KB
• Stars: 42
• Watchers: 3
• Forks: 12
• Open Issues: 9
• Metadata Files:
  • Readme: README.rst

Awesome Lists containing this project

• awesome-normalizing-flows - [Code

README

        
"Training Normalizing Flows with the Information Bottleneck for Competitive Generative Classification" (2020)

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

https://arxiv.org/abs/2001.06448

USAGE

^^^^^^^^^^^^^^

* All configuration files for all experiments in the paper are contained

  in the directory 'experiments_configs'.

* Training (for one example configuration):

  .. code:: sh

  

    python main.py train experiments_configs/cifar10/beta_ramp/beta_1.0000.ini

* Testing:

  .. code:: sh

   

    python main.py test experiments_configs/cifar10/beta_ramp/beta_1.0000.ini`

* The cifar/mnist datasets should be downloaded automatically the first time

  it is run. For the OoD evaluation, tiny imagenet and quickdraw have to be downloaded

  separately.

REQUIREMENTS

^^^^^^^^^^^^^^

To implement the INNs, we use of the FrEIA library

(github.com/VLL-HD/FrEIA)

**NOTE** This code currently only works with the previous 0.2 version of FrEIA. To install it:

.. code:: sh

    git clone https://github.com/VLL-HD/FrEIA.git

    cd FrEIA

    git checkout v0.2

    python setup.py develop

**Additional requirements:**

Python >= 3.6

.. code:: sh

    pip install -r requirements.txt