AntonioCarta/continual_cd_imm

Implementation of the Class-Domain Infinite Mixture Model from the paper "CD-IMM: The Benefits of Domain-based Mixture Models in Bayesian Continual Learning"

A Bayesian Non-Parametric Classifier for Continual Learning

CN-DPM and the Class VAE reuse the original code. Credits go to the original authors:

• CN-DPM
• Class VAE

The Dirichlet Process Mixture Models is a git submodule (available here).