Conference Agenda

Embracing the Tyranny of Testing

Moritz Hardt

Max Planck Institute for Intelligent Systems, Germany

Bridging Perception and Logic: An Abductive Learning Cycle for Semantically Anchored Facial Expression Recognition

Jens Johannes Gebele^1,2, Philipp Brune¹, Frank Schwab², Sebastian von Mammen²

1: Neu-Ulm University of Applied Sciences, Germany; 2: University of Würzburg

Human-in-the-loop Concept Discovery and Curation in Vision Foundation Models

Hyesu Lim¹, Jinho Choi², Jaegul Choo², Steffen Schneider¹

1: Helmholtz Munich, Germany; 2: KAIST AI, South Korea

TACTIC: Tabular-Attribute Conditioned Transformer for Image Classification

Laura Daza^1,2, Marta Hasny^1,2, Cristina González^1,2, Julia Schnabel^1,2,3

1: Helmholtz Munich, Germany; 2: Technical University of Munich, Germany; 3: King's College London, UK

Learning Physical Geometry from Noisy Helical Particle Tracks: A Comparative Study of Transformers, SBI, and JEPA

Sylvain Caillou, Jan Stark, Alexis Vallier, Catherine Biscarat

L2I Toulouse, CNRS/IN2P3, Université de Toulouse

MLflow pilot service for Helmholtz researchers

Lisana Berberi, Christophe Laures, Khadijeh Alibabaei, Valentin Kozlov, Achim Streit

Karlsruhe Institute of Technology (KIT), Germany

Towards Robust Foundation Models for Digital Pathology

Jonah Kömen^1,2, Edwin de Jong³, Julius Hense^1,2, Hannah Marienwald^1,2, Jonas Dippel^1,2,3, Philip Naumann^1,2, Eric Marcus⁴, Lukas Ruff³, Maximilian Alber^3,7, Jonas Teuwen⁴, Frederick Klauschen^1,5,6,7, Klaus-Robert Müller^1,2,8,9

1: Berlin Institute for the Foundations of Learning and Data (BIFOLD), Berlin, Germany; 2: Machine Learning Group, Technische Universität Berlin, Berlin, Germany; 3: Aignostics, Berlin, Germany; 4: The Netherlands Cancer Institute Amsterdam (NKI), Antoni van Leeuwenhoek Hospital (AvL), Amsterdam, Netherlands; 5: Institute of Pathology, Ludwig Maximilian University, Munich, Germany; 6: German Cancer Research Center, Heidelberg, and German Cancer Consortium, Munich, Germany; 7: Institute of Pathology, Charité Universitätsmedizin, Berlin, Germany; 8: Department of Artificial Intelligence, Korea University, Seoul, Korea; 9: Max-Planck Institute for Informatics, Saarbrücken, Germany

Sailing Past Syntax: A Human-in-the-Loop Framework for Safe Generative AI in Science

Nico Schuster

Centre de Physique des Particules de Marseille, France

The Helmholtz Model Zoo: Enabling AI Model Sharing and Inference in the Helmholtz Cloud

Hans Werners, Engin Eren, Philipp Heuser

Deutsches Elektronen-Synchroton DESY, Germany

Improving Reliability of LLM-Based Robotic Task Planning Through Domain Adaptation and Benchmarking

Kanak Pandit, Ananya Kulkarni, Muhammad Saeed

ARENA2036, Germany

Learning Actionable Insights from Scientific Data

Jilles Vreeken

CISPA Helmholtz Center for Information Security, Germany

RIBEX: Predicting and Explaining RNA Binding Across Structured and Intrinsically Disordered Regions (IDR)-rich Proteins

Samuele Firmani^1,2, Felix Steinbauer^2,4, Gjergji Kasneci^2,3,4, Annalisa Marsico¹, Marc Horlacher^1,2

1: Helmholtz Munich, Germany; 2: School of Computation, Information and Technology, TUM; 3: School of Social Sciences and Technology, TUM; 4: Munich Center for Machine Learning (MCML)

FoldSAE: Learning to Steer Protein Folding Through Sparse Representations

Wojciech Zarzecki^1,2,4, Paulina Szymczak³, Ewa Szczurek^1,3, Kamil Deja^2,4

1: University of Warsaw, Poland; 2: Warsaw University of Technology, Poland; 3: Helmholtz Munich, Germany; 4: Ideas Research Institute

Interpreting Multimodal Latent Spaces in Single-Cell Multi-omics with Feature Attribution Techniques

Anis Ismail, Alejandro Sifrim

Laboratory of Multi-omics Integrative Bioinformatics, Department of Human Genetics, KU Leuven, Leuven, Belgium.

Structured and interpretable patient embeddings from Single-Cell Foundation Models

Gonçalo Pinto

Helmholtz Munich, Germany

Modelling Urban Air Pollution with a Transferable Probabilistic Machine Learning Approach

Martin Ramacher¹, Paul Keil^1,2

1: Helmholtz Zentrum Hereon, Germany; 2: Helmholtz AI

Uncertainty-Guided Generation of Dark-Field Radiographs

Lina Felsner^1,2,3, Henriette Bast^4,5,6, Tina Dorosti^4,5,6, Florian Schaff^4,5,6, Franz Pfeiffer^4,5,6,7, Daniela Pfeiffer^5,6,7, Julia Schnabel^1,2,3,8

1: School of Computation, Information and Technology, Technical University of Munich, 85748 Garching, Germany; 2: Munich Center for Machine Learning (MCML), Munich, Germany; 3: Institute of Machine Learning in Biomedical Imaging, Helmholtz Munich, 85764 Neuherberg, Germany; 4: Department of Physics, School of Natural Sciences, Technical University of Munich, 85748 Garching, Germany; 5: Munich Institute of Biomedical Engineering, Technical University of Munich, 85748 Garching, Germany; 6: Institute for Diagnostic and Interventional Radiology, School of Medicine and Health, TUM University Hospital Klinikum rechts der Isar, Technical University of Munich, 81675 Munich, Germany; 7: Institute for Advanced Study, Technical University of Munich, 85748 Garching, Germany; 8: School of Biomedical Engineering & Imaging Sciences, King’s College, London, UK

SoraCT: Unconditioned 3D CT Synthesis via Video Diffusion Transformers

Jiayi Wang¹, Hadrien Reynaud¹, Bernhard Kainz^1,2

1: Friedrich-Alexander University Erlangen-Nürnberg, Germany; 2: Department of Computing, Imperial College London, UK

A 3D Foundation Model for Generalizable Biological Structure Segmentation in Tissue Clearing Images

Qi Liu^1,3,4,5, Ying Chen^1,3,4,5, Markus Elsner^1,6, Ali Ertürk^1,2,4,6,7

1: Institute for Stroke and Dementia Research, Klinikum der Universität München, Ludwig-Maximilians University Munich, Munich, Germany; 2: Institute for Intelligent Biotechnologies (iBIO), Helmholtz Center Munich, Neuherberg, Germany; 3: Faculty of Medicine, Ludwig-Maximilians University Munich, Munich, Germany; 4: Munich Cluster for Systems Neurology (SyNergy), Munich, Germany; 5: Munich Medical Research School (MMRS), Munich, Germany; 6: Deep Piction GmbH, Munich, Germany; 7: School of Medicine, Koç University, İstanbul, Turkey

Interpretable Representations for Hematology

Muhammed Furkan Dasdelen¹, Hyesu Lim¹, Paul Pommer¹, Michele Buck², Steffen Schneider¹, Carsten Marr¹

1: Helmholtz Munich, Germany; 2: Medical Department for Hematology and Oncology, Technical University Munich

Analysis of multi-omics data using graph neural networks identifies novel schizophrenia-associated genes suitable as drug targets

Svitlana Oleshko^1,2, Samuele Firmani¹, Jamie Robinson², Julien Hernandez-Lallement², Sergio Picart-Armada², Matthias Heinig¹, Till F.M. Andlauer^2,3, Annalisa Marsico²

1: Helmholtz Munich, Germany; 2: Boehringer Ingelheim Pharma, Germany; 3: Technical University of Munich, Germany

MLflow pilot service for Helmholtz researchers

Lisana Berberi, Christophe Laures, Khadijeh Alibabaei, Valentin Kozlov, Achim Streit

Karlsruhe Institute of Technology (KIT), Germany

Towards Robust Foundation Models for Digital Pathology

Jonah Kömen^1,2, Edwin de Jong³, Julius Hense^1,2, Hannah Marienwald^1,2, Jonas Dippel^1,2,3, Philip Naumann^1,2, Eric Marcus⁴, Lukas Ruff³, Maximilian Alber^3,7, Jonas Teuwen⁴, Frederick Klauschen^1,5,6,7, Klaus-Robert Müller^1,2,8,9

1: Berlin Institute for the Foundations of Learning and Data (BIFOLD), Berlin, Germany; 2: Machine Learning Group, Technische Universität Berlin, Berlin, Germany; 3: Aignostics, Berlin, Germany; 4: The Netherlands Cancer Institute Amsterdam (NKI), Antoni van Leeuwenhoek Hospital (AvL), Amsterdam, Netherlands; 5: Institute of Pathology, Ludwig Maximilian University, Munich, Germany; 6: German Cancer Research Center, Heidelberg, and German Cancer Consortium, Munich, Germany; 7: Institute of Pathology, Charité Universitätsmedizin, Berlin, Germany; 8: Department of Artificial Intelligence, Korea University, Seoul, Korea; 9: Max-Planck Institute for Informatics, Saarbrücken, Germany

Sailing Past Syntax: A Human-in-the-Loop Framework for Safe Generative AI in Science

Nico Schuster

Centre de Physique des Particules de Marseille, France

The Helmholtz Model Zoo: Enabling AI Model Sharing and Inference in the Helmholtz Cloud

Hans Werners, Engin Eren, Philipp Heuser

Deutsches Elektronen-Synchroton DESY, Germany

Improving Reliability of LLM-Based Robotic Task Planning Through Domain Adaptation and Benchmarking

Kanak Pandit, Ananya Kulkarni, Muhammad Saeed

ARENA2036, Germany

Modelling Urban Air Pollution with a Transferable Probabilistic Machine Learning Approach

Martin Ramacher¹, Paul Keil^1,2

1: Helmholtz Zentrum Hereon, Germany; 2: Helmholtz AI

Uncertainty-Guided Generation of Dark-Field Radiographs

Lina Felsner^1,2,3, Henriette Bast^4,5,6, Tina Dorosti^4,5,6, Florian Schaff^4,5,6, Franz Pfeiffer^4,5,6,7, Daniela Pfeiffer^5,6,7, Julia Schnabel^1,2,3,8

1: School of Computation, Information and Technology, Technical University of Munich, 85748 Garching, Germany; 2: Munich Center for Machine Learning (MCML), Munich, Germany; 3: Institute of Machine Learning in Biomedical Imaging, Helmholtz Munich, 85764 Neuherberg, Germany; 4: Department of Physics, School of Natural Sciences, Technical University of Munich, 85748 Garching, Germany; 5: Munich Institute of Biomedical Engineering, Technical University of Munich, 85748 Garching, Germany; 6: Institute for Diagnostic and Interventional Radiology, School of Medicine and Health, TUM University Hospital Klinikum rechts der Isar, Technical University of Munich, 81675 Munich, Germany; 7: Institute for Advanced Study, Technical University of Munich, 85748 Garching, Germany; 8: School of Biomedical Engineering & Imaging Sciences, King’s College, London, UK

SoraCT: Unconditioned 3D CT Synthesis via Video Diffusion Transformers

Jiayi Wang¹, Hadrien Reynaud¹, Bernhard Kainz^1,2

1: Friedrich-Alexander University Erlangen-Nürnberg, Germany; 2: Department of Computing, Imperial College London, UK