Skip to content

A variational inference method with accurate uncertainty estimation. It uses a new semi-implicit variational family built on neural networks and hierarchical distribution (ICML 2018).

License

Notifications You must be signed in to change notification settings

mingzhang-yin/SIVI

Repository files navigation

Semi-Implicit Variational Inference (SIVI)

Code to reproduce the results in Semi-Implicit Variational Inference.

Data sets

The data for SIVI_1d.py, SIVI_2d.py, SIVI_NB.py are self-generated in the python script.
The "waveform" data for SIVI_LR.py is in the data folder.
The MNIST data for SIVAE.py is self-contained.
Or of course, please try SIVI with your own datasets and probabilistic models.

(Notice in the code, the notation (J, K) is flipped with that in the paper.)

Citations

Below are the paper to cite if you find the algorithms in this repository useful in your own research:

@inproceedings{yin2018semi,
  title={Semi-Implicit Variational Inference},
  author={Yin, Mingzhang and Zhou, Mingyuan},
  booktitle={International Conference on Machine Learning},
  pages={5646--5655},
  year={2018}
}

License Info

This code is offered under the MIT License.

About

A variational inference method with accurate uncertainty estimation. It uses a new semi-implicit variational family built on neural networks and hierarchical distribution (ICML 2018).

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published