Skip to content

VAE optimized for reconstruction with the abiltiy to impute missing values

License

Notifications You must be signed in to change notification settings

rShar01/Imputation-Variational-Autoencoder

Repository files navigation

Class Based VAE

A ensemble model of VAE models. The dataset is partitioned by each label and a seperate VAE is trained on each partition. This allows the final ensemble model to account for sparse data better than traditional mixture VAEs. This allows the model to perform better on reconstruction tasks (even when the label is missing) and classification tasks.

See a more detailed write-up here

About

VAE optimized for reconstruction with the abiltiy to impute missing values

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published