Skip to content
Code for ICML 2019 paper on "Fast and Simple Natural-Gradient Variational Inference with Mixture of Exponential-family Approximations"
Branch: master
Clone or download
Latest commit a249b85 Jun 13, 2019
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
src/matlab add some comments Jun 7, 2019
README.md Update README.md Jun 13, 2019
poster_main.pdf update Jun 13, 2019
poster_workshop.pdf add a poster for gradient identities used in the paper Jun 7, 2019
report.pdf change the name of the file May 10, 2019

README.md

Code for ICML 2019 paper on Fast and Simple Natural-Gradient Variational Inference with Mixture of Exponential-family Approximations

  • To-do List:
    • added a poster about the gradient identities used in the paper (To appear at the ICML workshop on Stein's method) [done]
    • added a poster of the main paper [done]
    • [skewness] added Matlab code for the toy example using skew Gaussian and exponentially modified Gaussian [done]
    • [multi-modality] To add Matlab code for the toy example using MoG
    • [heavy tails] To add Matlab code for BLR using t-distribution and symmetric normal inverse Gaussian
    • To add Python code for Vadam extensions
You can’t perform that action at this time.