About me
I am a third-year PhD Candidate at ETH Zürich and the German Cancer Research Center (DKFZ), also affiliated with Helmholtz Imaging. My work is dedicated to advancing scientific discovery through explainable AI (XAI) applied to high-dimensional and multi-modal data. I contributed to multiple interdisciplinary discovery projects across the domains of 🩻 medical imaging, 🦠 microbiology, and 🧱 material sciences. My efforts extend to increasing the trustworthiness of XAI results by evaluating methods and metrics, particularly focusing on attention and attribution methods.
🎓 Academic Education
- B.Sc. in Economics at University of Mannheim
- M.Sc. in Statistics at University of Geneva
- Exchange, Master Thesis & Research Intern at ETH Zürich
- (Current) Ph.D. in Informatics in the IML Group DKFZ & ISE Group ETH Zürich
- (Current) DKFZ International PhD Program & Fellowship at the Helmholtz International Graduate School for Cancer Research
📑 Publications
Paper (Published) | Ressources |
---|---|
Discovering Process Dynamic for Scalable Perovskite Solar Cell Manufacturing with Explainable AIAuthors: Lukas Klein, Sebastian Ziegler, Felix Laufer, Charlotte Debus, Markus Götz, Klaus Maier-Hein, Ulrich W. Paetzold, Fabian Isensee, Paul F. JaegerJournal: Advanced Materials (IF: 30) Date: 12.2023 |
|
Understanding Solar Cell Manufacturing with Explainable AIAuthors: Lukas Klein, Sebastian Ziegler, Felix Laufer, Charlotte Debus, Markus Götz, Klaus Maier-Hein, Ulrich W. Paetzold, Fabian Isensee, Paul F. JaegerVenue: NeurIPS 2023 XAIA Workshop (Oral, Top 4 Paper) Date: 12.2023 |
|
Navigating the Pitfalls of Active Learning Evaluation: A Systematic Framework for Meaningful Performance AssessmentAuthors: Carsten T. Lüth, Till J. Bungert, Lukas Klein, Paul F. JaegerVenue: NeurIPS 2023 Date: 12.2023 |
|
A Call to Reflect on Evaluation Practices for Failure Detection in Image ClassificationAuthors: Paul F. Jaeger, Carsten T. Lueth, Lukas Klein, Till J. BungertVenue: ICLR 2023 (Oral) Date: 05.2023 |
|
From Correlation to Causation: Formalizing Interpretable Machine Learning as a Statistical ProcessAuthors: Lukas Klein, Mennatallah El-Assady, Paul F JaegerVenue: IJCAI 2022 Workshop on XAI Date: 07.2022 |
|
Improving Explainability of Disentangled Representations using Multipath-Attribution MappingsAuthors: Lukas Klein, João B. S. Carvalho, Mennatallah El-Assady, Paolo Penna, Joachim M. Buhmann, Paul F JaegerVenue: MIDL 2022 Date: 04.2022 |
|
DENKIMPULS DIGITALE ETHIK: Einsatz von KI in der WirtschaftAuthors: Astrid Aupperle, Thomas Langkabel, Lukas Klein (Microsoft Germany GmbH)Venue: Initiative D21: Ethics Working Group Date: 01.2018 |
Paper (Under Review) | Ressources |
---|---|
Anonymous Title: Scientific Discovery in Diabetes-type 2 through Multi-Modal Light-and Fluorescence Microscopy Data.Authors: Lukas Klein, Sebastian Ziegler, Yanni Morgenroth, Eyke Schoeniger, Michele Solimena, Felicia Gerst, Robert Wagner, Fabian Isensee, Klaus Maier-Hein, Paul F. JaegerVenue: Under review |
|
Anonymous Title: Attribution & Attention Method and Metric EvaluationAuthors: Lukas Klein, Carsten Lueth, Udo Schlegel, Till Bungert, Mennatallah El-Assady, Paul F.JaegerVenue: Under review |
|
Anonymous Title: Predictive Imaging Biomarker Discovery for Treatment Effect AnalysisAuthors: Shuhan Xiao, Lukas Klein, Paul F. Jaeger, Jens Petersen, Klaus Maier-Hein,Venue: Under review |
|
📨 Contact
-
Interactive Machine Learning Group (E290)
Deutsches Krebsforschungszentrum
Im Neuenheimer Feld 223
69120 Heidelberg - lukas.klein(at)dkfz.de
- GitHub
- Google Scholar