Model-based framework for robust classification that jointly accounts for outliers, label noise and unobserved classes in the test set, employing a MVN mixture model with Parsimonious structure.
This repository is associated with the paper Cappozzo, Greselin, Murphy (2020) Anomaly and Novelty detection for robust semi-supervised learning. https://doi.org/10.1007/s11222-020-09959-1
You can install the development version of raedda from github with:
devtools::install_github("AndreaCappozzo/raedda")