https://github.com/djsutherland/opt-mmd

Learning kernels to maximize the power of MMD tests
https://github.com/djsutherland/opt-mmd

generative-adversarial-network hypothesis-testing kernel-methods machine-learning maximum-mean-discrepancy python shogun statistical-tests tensorflow theano

Last synced: about 1 month ago
JSON representation

Learning kernels to maximize the power of MMD tests

• Host: GitHub
• URL: https://github.com/djsutherland/opt-mmd
• Owner: djsutherland
• License: bsd-3-clause
• Created: 2016-11-17T22:23:29.000Z (about 8 years ago)
• Default Branch: master
• Last Pushed: 2018-01-11T16:34:15.000Z (about 7 years ago)
• Last Synced: 2024-12-12T21:51:27.600Z (about 2 months ago)
• Topics: generative-adversarial-network, hypothesis-testing, kernel-methods, machine-learning, maximum-mean-discrepancy, python, shogun, statistical-tests, tensorflow, theano
• Language: Python
• Homepage: https://arxiv.org/abs/1611.04488
• Size: 97.7 KB
• Stars: 206
• Watchers: 13
• Forks: 73
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
Code for the paper "Generative Models and Model Criticism via Optimized Maximum Mean Discrepancy" ([arXiv:1611.04488](https://arxiv.org/abs/1611.04488); [published at](https://openreview.net/forum?id=HJWHIKqgl) ICLR 2017), by [Dougal J. Sutherland](http://www.gatsby.ucl.ac.uk/~dougals/) ([@dougalsutherland](https://github.com/dougalsutherland)), [Hsiao-Yu Tung](http://sfish0101.bitbucket.io/), [Heiko Strathmann](http://herrstrathmann.de/about/) ([@karlnapf](https://github.com/karlnapf)), Soumyajit De ([@lambday](https://github.com/lambday)), [Aaditya Ramdas](https://people.eecs.berkeley.edu/~aramdas/), [Alex Smola](https://alex.smola.org/), and [Arthur Gretton](http://www.gatsby.ucl.ac.uk/~gretton/).

- Implementations of the variance estimator are in Theano in [`two_sample/mmd.py`](two_sample/mmd.py) and in Tensorflow in [`gan/mmd.py`](gan/mmd.py).

- General code for learning kernels for a fixed two-sample test, with Theano, is in [two_sample](two_sample).

- Code for the GAN variants, using TensorFlow, is in [gan](gan).

- Code for the efficient permutation test described in Section 3 is in the 6.0 release of [Shogun](http://shogun.ml); look under [`shogun/src/shogun/statistical_testing`](https://github.com/shogun-toolbox/shogun/tree/develop/src/shogun/statistical_testing). An example of using it in the Python API is in [`two_sample/mmd_test.py`](two_sample/mmd_test.py).

This code is under a BSD license, but if you use it, please cite the paper.