Workshop contribution MLJ

Diego Arenas, Edoardo Barp, Gergö Bohner, Valentin Churvay, Franz Kiraly, Thibaut Lienart, Sebastian Vollmer, Mike Innes, Bernd Bischl

Keywords: machine learning toolbox, automl, workflow, meta package

TL;DR: Task based machine learning tool box like MLR - but in Julia

Abstract: MLJ is a machine learning toolbox for Julia with the aim to provide a coherent interface to run and compare machine learning models. The software should support basic supervised learning framework implementing key parts of workflow for most important modelling pipelines including tuned, ensembles, and deep learning/neural networks (via interfacing). Further development will extend to cover advanced ML tasks and workflows. This project is building an MLR type machine learning toolbox in Julia.

Decision: reject