DataNinja Spring School 2023 Online Proceedings

Welcome to the online proceedings of the DataNinja Spring School 2023 – “Trustworthy AI: Building Safe and Reliable Solutions”.
During our Spring School, young talents in AI had the opportunity to showcase their own work in the virtual poster session.
This page serves as a repository for the corresponding abstracts and poster files.


Unsupervised Neural Network Verification

Benedikt Böing1 & Emmanuel Müller1
1 Chair of Data Science and Data Engineering, Technical Uni- versity of Dortmund, Dortmund, Germany


Continual Hyperband

Jasmin Brandt1, Marcel Wever2, Dimitrios Iliadis3, Viktor Bengs2 & Eyke Hüllermeier2
1 University of Paderborn, Paderborn, Germany
2 University of Munich, Munich, Germany
3 Ghent University, Ghent, Belgium


Provably Bounding Neural Network Preimages

Suhas Kotha1, Christopher Brix2, Zico Kolter1,4, Krishnamurthy (Dj) Dvijotham3 & Huan Zhang1
1 Carnegie Mellon University, Pittsburgh PA, USA
2 RWTH Aachen University, Aachen, Germany
3 Google Research, Brain Team
4 Bosch Center for AI


An Investigation of the Vulnerabilities and Effects of Prompt-Tuning
to Pre-Trained Language Models

Fotini Deligiannaki1 & Arne Peter Raulf1
1 Institute for AI Safety and Security, German Aerospace Center, Germany


Finding the Relevant Samples for Decision Trees in Reinforcement Learning

Raphael C. Engelhardt1, Moritz Lange2, Laurenz Wiskott2 & Wolfgang Konen1
1 Cologne Institute of Computer Science, TH Köln, Köln, Germany
2 Institute for Neural Computation, Ruhr-University Bochum, Bochum, Germany


Graph Learning by Dynamic Sampling

Luca Hermes1, Aleksei Liuliakov1 & Malte Schilling2
1 Machine Learning Group, Bielefeld University, Bielefeld, Germany
2 Autonomous Intelligent Systems Group, University of Münster, Münster, Germany


Unsupervised DeepView: Global Explainability of Uncertainties for High Dimensional Data

Carina Newen1 & Emmanuel Müller1
1 Research Center Trustworthy Data Science and Security, Technical University of Dortmund, Dortmund, Germany


Identifying and Explaining Undesirable Traces in Business Processes Using Descriptive and Predictive Analysis

Ali Norouzifar1
1 Process and Data Science Group (PADS), RWTH Aachen University, Aachen, Germany


LU-Net: Invertible Neural Networks Based on Matrix Factorization

Robin Chan1, Sarina Penquitt2 & Hanno Gottschalk2
1 Machine Learning Group, Bielefeld University, Bielefeld, Germany
2 IZMD, University of Wuppertal, Wuppertal, Germany


Off-line Learning Analysis for Soft Committee Machines with GELU Activation

Frederieke Richert1, Michiel Straat2 & Michael Biehl1
1 Bernoulli Institute for Mathematics, Computer Science and Artificial Intelligence, University of Groningen, Groningen, The Netherlands
2 Center for Cognitive Interaction Technology, Bielefeld University, Bielefeld, Germany


Evaluating Spiking Neural Network Models: A Comparative Performance Analysis

Sanaullah1, Shamini Koravuna2, Ulrich Rückert2 & Thorsten Jungeblut1
1 Bielefeld University of Applied Science, Bielefeld, Germany
2 University of Bielefeld, Bielefeld, Germany