Hands-On: Auto-Sklearn

Abstract

TL;DR: ML in 4 lines of Python!

Do you want to learn how to use our popular OSS AutoML tool Auto-sklearn? Then this hands-on is right for you. You will learn how to move from manually constructing scikit-learn ML pipelines to full AutoML using the popular open-source Auto-sklearn library – a drop-in replacement for any scikit-learn estimator. We will start with a short presentation, and then you will, in a group, work through several interactive tasks where you learn how to apply, modify and understand Auto-sklearn to be ready to use it for your own projects afterwards.

Bio

Matthias Feurer is a doctoral candidate at the Machine Learning Lab at the University of Freiburg, Germany. His research focuses on automated machine learning, hyperparameter optimization and Meta-Learning. He is actively involved in developing open source software for AutoML and is the maintainer and founder of Auto-sklearn and OpenML-Python. Matthias is a founding member of the Open Machine Learning Foundation, gave AutoML tutorials at the GCPR and ECMLPKDD summer school and an invited talk at the AutoML workshop in 2021. Furthermore, he co-organized the AutoML workshops in 2019 and 2020, the 1st AutoML fall school in 2021, and was one of the social chairs of the 1st AutoML conference in 2022.

Katharina Eggensperger is a PostDoc in the Machine Learning Lab at the University of Freiburg, Germany, where she also defended her PhD in 2022. Her interests focus on empirical performance modeling, automated machine learning and hyperparameter optimization. With her research, she aims to make machine learning easy-to-use, as demonstrated by the successful projects Auto-sklearn and SMAC3 to which she is contributing. She has been an invited speaker at the BayesOpt workshop at NeurIPS 2016, co-organized the AutoML workshop series at ICML from 2019 until 2021 and was a co-organizer of the 1st AutoML conf in 2022, as well as of the AutoML fall school in 2021.

Eddie Bergman is a Research Engineer at the Machine Learning Lab at the University of Freiburg, Germany. He is passionate about making the forefront of AutoML research accessible to everyone and helping researchers through robust, tested and well designed tools. His main focus is working on the group's stack of widely adopted tools such as AutoSklearn, SMAC and ConfigSpace. He took part in the AutoML Fall School last year as a participant and hopes to create the best experience for students having been on both sides of the coin.