Towards Explainable Automated Machine Learning

MCML Authors

Julia Moosbauer

Dr.

* Former Member

→ Group Bernd Bischl
 Statistical Learning and Data Science

Abstract

This thesis explores the intersection of Automated Machine Learning (AutoML) and explainable AI, addressing the need for transparency at multiple levels: the model, the learning algorithm, and the AutoML system itself. The work develops methods for enhancing model explainability through multi-objective hyperparameter optimization (HPO) and introduces new techniques to understand the effects of hyperparameters and optimizers within AutoML systems. These contributions advance the field by providing more interpretable and reliable tools for AutoML, ultimately increasing the accessibility and trustworthiness of machine learning models and their deployment. (Shortened.)

phdthesis

---

Dissertation

LMU München. May. 2023

---

Authors

J. Moosbauer

---

Links

DOI

---

Research Area

 A1 | Statistical Foundations & Explainability