Skip to content

Latest commit

 

History

History
12 lines (8 loc) · 506 Bytes

README.md

File metadata and controls

12 lines (8 loc) · 506 Bytes

not-MIWAE

Code accompanying the paper
Niels Bruun Ipsen, Pierre-Alexandre Mattei, and Jes Frellsen.
not-MIWAE: Deep generative modelling with missing not at random data.
arXiv preprint arXiv:2006.12871 (2020).

Shows how to learn deep generative models with missing data under the MNAR assumption.
The notebook not-MIWAE-demo.ipynb introduces the model and training step by step.
task01.py runs the not-MIWAE and competitors on a UCI dataset.