Conference Agenda
Overview and details of the sessions of this conference. Please select a date or location to show only sessions at that day or location. Please select a single session for detailed view (with abstracts and downloads if available).
Please note that all times are shown in the time zone of the conference. The current conference time is: 13th June 2026, 12:50:40pm CEST
|
Daily Overview |
| Session | ||
Poster Session II
| ||
| Session Abstract | ||
|
Discover cutting-edge AI research and innovations at the Poster Session. Connect with authors, ask questions, and engage in lively discussions that spark collaboration and new ideas. | ||
    | ||
ID: 243
/ Poster No. # 10: 001
Modalities: Image Methods: Other Application Domain: Health HemAutomaton: A lightweight NCA-based pipeline for large-scale extraction of white blood cells from whole slide images 1Helmholtz Munich, Germany; 2Universitätsklinikum Erlangen, Germany ID: 290
/ Poster No. # 10: 002
Modalities: Other Methods: Foundation Models Application Domain: Health DeepRVAT2: Unified Modeling of Coding and Regulatory Rare Variation at Genome Scale for Enhanced Gene Discovery and Diagnostics 1Computational Health Center, Helmholtz Center Munich, Neuherberg, Germany; 2chool ofComputation, Information and Technology, Technical University of Munich, Garching, Germany; 3Institute of Human Genetics, School of Medicine, Technical University of Munich, Munich, Germany; 4German Cancer Research Center (DKFZ), Heidelberg, Germany; 5European Molecular Biology Laboratory (EMBL), Heidelberg, Germany; 6Heidelberg University, Germany ID: 1215
/ Poster No. # 10: 003
Modalities: Image Methods: Foundation Models, Other Application Domain: Health The Mean is the Mirage: Entropy-Adaptive Model Mergingunder Heterogeneous Domain Shifts in Medical Imaging 1School of Computation, Information and Technology, Technical University of Munich, Germany; 2Institute of Machine Learning in Biomedical Imaging, Helmholtz Munich, Germany; 3relAI – Konrad Zuse School of Excellence in Reliable AI; 4Munich Center for Machine Learning (MCML); 5Institute of Pathology, Technical University of Munich, Germany; 6School of Biomedical Engineering and Imaging Sciences, King’s College London, UK. ID: 1145
/ Poster No. # 10: 004
Modalities: Tabular Data Methods: Other Application Domain: Health ConvexGating infers gating strategies from clusters in single cell cytometry data 1University of Leipzig, Institute for Medical Informatics, Statistics, and Epidemiology, Leipzig, Germany; 2Center for Scalable Data Analytics and Artificial Intelligence (ScaDS.AI), Leipzig, Germany; 3Systems Medicine, Deutsches Zentrum für Neurodegenerative Erkrankungen (DZNE), Bonn, Germany; 4Modular High Performance Computing and Artificial Intelligence, German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany; 5Research Group Tissue Control of Immunocytes, Helmholtz Center Munich, Munich, Germany; 6Life and Medical Sciences (LIMES) Institute, University of Bonn, Bonn, Germany; 7PRECISE Platform for Single Cell Genomics and Epigenomics, DZNE and University of Bonn and West German Genome Center (WGGC), Bonn, Germany; 8Immunogenomics & Neurodegeneration, Deutsches Zentrum für Neurodegenerative Erkrankungen (DZNE), Bonn, Germany; 9Department of Microbiology and Immunology, The Peter Doherty Institute for Infection and Immunity, University of Melbourne, Melbourne, VIC, Australia; 10Molecular Immunology in Neurodegeneration, German Center for Neurodegenerative Diseases and the University of Bonn, Germany; 11Medical Microbiology and Immunology Department, Faculty of Medicine, Mansoura University, Egypt; 12Institute of Innate Immunity, Biophysical Imaging, Medical Faculty, University of Bonn, Bonn, Germany; 13Max Planck Institute for Metabolism Research, Center for Endocrinology, Diabetes and Preventive Medicine (CEDP), Cologne, Germany; 14University of Leipzig, Faculty of Mathematics and Computer Science, Leipzig, Germany; 15Institute of Computational Biology, Helmholtz Center Munich, Germany; 16Department of Mathematics, Technical University of Munich, Germany; 17TUM School of Life Sciences Weihenstephan, Technical University of Munich, Germany ID: 1278
/ Poster No. # 10: 005
Modalities: Multimodal Data Methods: Foundation Models Application Domain: Earth & Environment Cross Modalities Pretraining of Sparse Lidar and Dense Image Foundation Model for Global Carbon Stock Mapping 1Helmholtz-Zentrum Dresden-Rossendorf, Dresden, Germany; 2Chair of Data Science in Earth Observation, Technical University of Munich, Munich, Germany ID: 1151
/ Poster No. # 10: 006
Modalities: Tabular Data, Text Methods: Agentic AI, Uncertainty Quantification Application Domain: Health ICD-Code Extraction from Clinical Notes using Large Language Models in a RAG pipeline 1Hybrid Methods in Artificial Intelligence and Machine Learning, University of Rostock, Germany; 2German Center for Neurodegenerative Diseases, Rostock, Germany ID: 1225
/ Poster No. # 10: 007
Modalities: Tabular Data Methods: Other Application Domain: Health Causal Machine Learning for Predictive Biomarker Discovery and Subgroup Refinement in Metastatic Colorectal Cancer 1Department of Biochemistry and Pharmacology, Bio21 Molecular Science and Biotechnology Institute, The University of Melbourne, Australia; 2Department of Medicine III and Comprehensive Cancer Center Munich, University Hospital, Ludwig-Maximilians University Munich, Germany; 3Comprehensive Cancer Center Munich, Germany; 4German Cancer Consortium (DKTK), partner site Munich, German Cancer Research Center (DKFZ), Germany; 5LMU Munich School of Management, LMU Munich, Germany; 6Munich Center for Machine Learning, Germany; 7Computational Health Center, Institute of Computational Biology, Helmholtz Munich, Germany ID: 1187
/ Poster No. # 10: 008
Modalities: Simulation Data, Tabular Data, Time Series Methods: Physics-informed Machine Learning Application Domain: Matter Neural Operator-Based Surrogate Modeling for Efficient Prediction of Temperature and Residual Stresses in Tempered Glass Universität Augsburg, Germany ID: 1368
/ Poster No. # 10: 009
Modalities: Graphs, Other Methods: Graph Neural Networks, Physics-informed Machine Learning Application Domain: Earth & Environment, Health GRIP: Physics-Informed Neural Network for Gradient Retention Time Prediction in Liquid Chromatography 1Helmholtz Institute for Pharmaceutical Research Saarland (HIPS), Helmholtz Centre for Infection Research (HZI), Germany; 2German Research Center for Artificial Intelligence (DFKI) Kaiserslautern ID: 1252
/ Poster No. # 10: 010
Modalities: Text Methods: Foundation Models, Other Application Domain: Core Machine Learning, Information Who Owns Human Experience? Ethical Implications of Transforming Tacit Knowledge into Neural Models 1University Augsburg, Germany; 2ergonoi GbR ID: 1325
/ Poster No. # 10: 011
Modalities: Tabular Data, Other Methods: Generative Models, Probabilistic Methods Application Domain: Health Towards Useful and Private Synthetic Omics: Community Benchmarking of Generative Models for Transcriptomics Data 1European Molecular Biology Laboratory (EMBL), Genome Biology Unit, Heidelberg, Germany; 2Division of Computational Genomics and Systems Genetics, German Cancer Research Center (DKFZ), Heidelberg, Germany; 3CISPA Helmholtz Center for Information Security, Saarbrücken, Germany; 4University of Helsinki, Finland; 5Heidelberg University, Germany; 6Helmholtz Munich, Germany; 7Division of Tumorigenesis and Molecular Cancer Prevention, German Cancer Research Center (DKFZ), Heidelberg, Germany; 8DKFZ Hector Cancer Institute at the University Medical Center Mannheim, Germany; 9Eberhard Karls Universität Tübingen, Germany; 10University of Washington Tacoma, USA; 11Sage Bionetworks, Seattle, USA; 12Ghent University, Ghent, Belgium; 13European Bioinformatics Institute (EMBL-EBI), UK ID: 372
/ Poster No. # 10: 012
Modalities: Graphs, Multimodal Data, Tabular Data, Text Methods: Uncertainty Quantification, Other Application Domain: Core Machine Learning, Health Multimodal Representation Learning for Pan-Cancer Type Classification Using Attention-Based Set Encoding and Contrastive Pre-Training Max Delbrück Center for Molecular Medicine, Germany ID: 175
/ Poster No. # 10: 013
Modalities: Simulation Data, Time Series Methods: Probabilistic Methods, Uncertainty Quantification Application Domain: Core Machine Learning Higher-Order Hit-&-Run Samplers for Linearly Constrained Densities 1Forschungszentrum Jülich, Germany; 2Department of Statistics, LMU Munich; 3Munich Center for Machine Learning (MCML) ID: 251
/ Poster No. # 10: 014
Modalities: Multimodal Data, Tabular Data Methods: Probabilistic Methods, Other Application Domain: Core Machine Learning, Health Simple Yet Effective: Basic Population and Evolutionary Statistics Contain Richest Information to Infer Systematic Variant Effects 11Institute of Translational Genomics, Helmholtz Zentrum München, German Research Center for Environmental Health, Neuherberg, Germany; 2School of Computation, Information and Technology, Technical University of Munich, Garching, Germany; 3Helmholtz Association—Munich School for Data Science (MUDS), Munich, Germany; 4Helmholtz Pioneer Campus, Helmholtz Zentrum München, German Research Center for Environmental Health, Neuherberg, Germany; 5Department of Pediatrics, Dr. von Hauner Children's Hospital, University Hospital, Ludwig Maximilian University Munich, Munich, Germany ID: 351
/ Poster No. # 10: 015
Modalities: Text Methods: Probabilistic Methods, Uncertainty Quantification, Other Application Domain: Earth & Environment Bringing Gaps between Farmers and AI with XAI Principles 1German Aerospace Center (DLR), Germany; 2German Aerospace Center (DLR), Germany; 3German Aerospace Center (DLR), Germany ID: 322
/ Poster No. # 10: 016
Modalities: Text Methods: Agentic AI, Generative Models Application Domain: Information Abstracts Explorer: LLM-Powered Semantic Exploration of Scientific Conference Proceedings Helmholtz Zentrum Dresden Rossendorf (HZDR), Germany ID: 332
/ Poster No. # 10: 017
Modalities: Multimodal Data Methods: Uncertainty Quantification, Other Application Domain: Core Machine Learning Systematic benchmarking of imaging-based spatial transcriptomics preprocessing 1Institute of Computational Biology, Helmholtz Munich, Germany; 2Institute of Lung Health and Immunity, Helmholtz Munich, Germany; 3Indiana University School of Medicine ID: 191
/ Poster No. # 10: 018
Modalities: Image, Multimodal Data, Text Methods: Agentic AI, Uncertainty Quantification Application Domain: Health Dynamic Decision Learning: Test-Time Evolution for Abnormality Grounding in Rare Diseases 1Technical University of Munich, Germany; 2University of Trento, Italy; 3Imperial College London, United Kingdom; 4Helmholtz Munich, Germany; 5King's College London, United Kingdom ID: 177
/ Poster No. # 10: 019
Modalities: Graphs Methods: Graph Neural Networks Application Domain: Core Machine Learning Geometry-Aware Edge Pooling for Graph Neural Networks 1Helmholtz Munich, Germany; 2Technical University of Munich, Germany; 3University of Montreal, Canada; 4Mila - Quebec Artificial Intelligence Institute, Canada; 5University of Fribourg, Switzerland ID: 157
/ Poster No. # 10: 020
Modalities: Multimodal Data Methods: Foundation Models Application Domain: Health Phenotype-driven virtual screening accelerates therapeutic discovery for idiopathic pulmonary fibrosis 1Helmholtz Munich; 2Faculty of Mathematics, Informatics and Mechanics, University of Warsaw ID: 347
/ Poster No. # 10: 021
Modalities: Tabular Data, Other Methods: Generative Models, Other Application Domain: Health Machine Learning-based RNA Design for the Development of Efficient mRNA Therapeutics Helmholtz Munich, Computational Health Center, Computational Biology (ICB), Germany ID: 328
/ Poster No. # 10: 022
Modalities: Text Methods: Other Application Domain: Core Machine Learning Reinterpreting Protein Sequences as Structured Signals: A CNN-Residual Approach to Neuropeptide Classification 1Federal University of Technology-Paraná, Brazil; 2Helmholtz Centre for Environmental Research, Deutschland ID: 333
/ Poster No. # 10: 023
Modalities: Image, Multimodal Data, Other Methods: Other Application Domain: Aeronautics, Space & Transport Extracting 1D JWST Spectra from 2D Spectrograms with Deep Learning 1Centre de Physique des Particules de Marseille, France; 2Laboratoire d'Informatique et des Systèmes, Marseille, France ID: 204
/ Poster No. # 10: 024
Modalities: Image Methods: Uncertainty Quantification Application Domain: Health Segmentation of glands in big cohort data by deep learning Universitätsmedizin Greifswald, Germany ID: 198
/ Poster No. # 10: 025
Modalities: Image Methods: Generative Models Application Domain: Health Subclass-Conditioned Flow Matching for Long-Tailed Chest X-Ray Augmentation 1Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), Germany; 2Imperial College London, UK ID: 376
/ Poster No. # 10: 026
Modalities: Image Methods: Foundation Models, Other Application Domain: Health MoA: Mixture of Aggregators Improves Slide-Level Diagnosis in Computational Pathology 1Helmholtz Munich, Germany; 2TUM; 3LMU ID: 231
/ Poster No. # 10: 027
Modalities: Text Methods: Other Application Domain: Aeronautics, Space & Transport Topic Modeling and Opinion Analysis of Hyperloop Discussions on Twitter Using Latent Dirichlet Allocation 1University of Stuttgart , Germany; 2University of Southern California, USA; 3Boston University, USA; 4ARENA2036 e.V.; 5University of Mumbai, India; 6Hochschule Bremen, Germany; 7Swinburne University of Technology, Melbourne, Australia ID: 229
/ Poster No. # 10: 028
Modalities: Text Methods: Other Application Domain: Aeronautics, Space & Transport Sentiment Analysis of Indian Hyperloop Tweets for Opinion Inversion Prediction 1University of Stuttgart, Germany; 2University of Southern California, USA; 3Boston University, USA; 4ARENA2036, Germany; 5Deakin University; 6Swinburne University of Technology, Australia; 7University of Mumbai, India ID: 268
/ Poster No. # 10: 029
Modalities: Image, Simulation Data Methods: Physics-informed Machine Learning, Other Application Domain: Health Test-time augmentation with synthetic data addresses distribution shifts in spectral imaging 1Division of Intelligent Medical Systems (IMSY), German Cancer Research Center (DKFZ), Heidelberg, Germany; 2Helmholtz Information and Data Science School for Health, Karlsruhe/Heidelberg, Germany; 3Faculty of Mathematics and Computer Science, Heidelberg University, Heidelberg, Germany; 4Department of General, Visceral, and Transplantation Surgery, Heidelberg University Hospital, Heidelberg, Germany; 5National Center for Tumor Diseases (NCT), A Partnership between DKFZ and University Medical Center Heidelberg, Heidelberg, Germany; 6Department of Urology, University Medical Center Mannheim, Heidelberg University, Mannheim, Germany; 7Medical Faculty, Heidelberg University, Heidelberg, Germany ID: 241
/ Poster No. # 10: 030
Modalities: Image Methods: Generative Models, Physics-informed Machine Learning Application Domain: Core Machine Learning HoloZip: A Machine Learning-Based Approach for Extreme X-ray Holographic Data Compression 1Deutsches Elektronen-Synchrotron DES, Germany; 2Department Physik, Universität Hamburg UHH, Germany; 3HELMHOLTZ IMAGING ID: 183
/ Poster No. # 10: 032
Modalities: Audio, Graphs, Image, Multimodal Data, Simulation Data, Tabular Data, Text Methods: Probabilistic Methods Application Domain: Core Machine Learning Optimal conversion from Rényi Differential Privacy to $f$-Differential Privacy 1Helmholtz Munich, Germany; 2Technical University of Munich, Germany; 3Harvard University, USA; 4Hasso-Plattner-Institut, Germany ID: 260
/ Poster No. # 10: 033
Modalities: Image, Multimodal Data, Text Methods: Foundation Models Application Domain: Core Machine Learning SOTAlign: Semi-Supervised Alignment of Unimodal Vision and Language Models via Optimal Transport 1Helmholtz Munich, Germany; 2Technical University of Munich, Germany; 3Munich Center for Machine Learning, Germany; 4Munich Data Science Institute, Germany; 5Télécom Paris, France; 6École Polytechnique, France ID: 330
/ Poster No. # 10: 034
Modalities: Other Methods: Probabilistic Methods, Uncertainty Quantification Application Domain: Health AtlasAlign: PointTransformer-based alignment of spatial omics to mouse brain common coordinate framework 1Technical University of Munich (TUM), Germany; 2Institute for Stroke and Dementia Research (ISD), Germany; 3Institute of Neuroscience and Medicine (INM-1) Forschungszentrum Jülich, Germany; 4Institute of Computational Biology, Helmholtz Munich, Germany ID: 158
/ Poster No. # 10: 035
Modalities: Other Methods: Foundation Models, Physics-informed Machine Learning Application Domain: Health UdonPred: Untangling Protein Intrinsic DisorderPrediction 1School of Computation, Information and Technology (CIT), Faculty of Informatics, Chair of Bioinformatics & Computational Biology - i12, Technical University of Munich (TUM), Boltzmannstr. 3, 85748 Garching/Munich, Germany; 2Institute of Computational Biology (ICB), Helmholtz München, Ingolstädter Landstraße 1, 85764 Neuherberg, Germany; 3Translational Microbiome Data Integration, School of Life Sciences, Technical University of Munich (TUM), Alte Akademie 8, 85354 Freising, Germany; 4Institute for Advanced Study (TUM-IAS), Technical University of Munich (TUM), Lichtenbergstr. 2a, 85748 Garching/Munich, Germany; 5School of Life Sciences Weihenstephan (TUM-WZW), Technical University of Munich (TUM), Alte Akademie 8, 85354 Freising, Germany ID: 254
/ Poster No. # 10: 036
Modalities: Text Methods: Foundation Models Application Domain: Core Machine Learning Fused Transformer Blocks Improve Pretraining Efficiency and Contextual Language Understanding factorize.bio, Germany ID: 396
/ Poster No. # 10: 037
Persona Vectors: Steering Towards Faithful Verbalized Uncertainty Expression in LLMs 1Helmholtz AI, Munich; 2Technical University of Munich; 3University of Toronto; 4Technical University of Nuremberg ID: 131
/ Poster No. # 10: 038
Modalities: Image, Multimodal Data, Text Methods: Other Application Domain: Core Machine Learning Scalable and interpretable representation alignment with ordinal similarity 1Helmholtz Munich, AIH, Germany; 2Technical University of Munich, Germany; 3University of Warsaw, Poland ID: 244
/ Poster No. # 10: 039
Modalities: Multimodal Data Methods: Reinforcement Learning Application Domain: Earth & Environment Multimodal Data-driven Transfer Learning Impelled Framework for Spatiotemporal Forecasting of Urban Pollution (PM2.5) Research Institute for Sustainability, Germany ID: 166
/ Poster No. # 10: 040
Modalities: Other Methods: Probabilistic Methods Application Domain: Health A Hierarchical Bayesian Multiple Instance Learning Framework for Joint Localization and Testing of Spatial Differential Expression 1Helmholtz Munich, Germany; 2Technical University of Munich, Germany; 3Regensburg University, Germany; 4ETH Zurich, Switzerland ID: 355
/ Poster No. # 10: 043
Modalities: Image, Simulation Data Methods: Probabilistic Methods, Uncertainty Quantification Application Domain: Information Trustworthy Deep Surrogates for Inverse Materials Design with Calibrated Uncertainty 1Institute for Advanced Simulations - Materials Data Science and Informatics (IAS-9), Forschungszentrum Jülich GmbH; 2Chair of Materials Data Science and Materials Informatics, Faculty 5 – Georesources and Materials Engineering, RWTH Aachen University ID: 307
/ Poster No. # 10: 044
Modalities: Image Methods: Reinforcement Learning, Other Application Domain: Health In-Model Bias Mitigation Methods for Age Confounding in Dementia Detection 1Deutsches Zentrum für Neurodegenerative Erkrankungen (DZNE), Rostock, Germany; 2Klinik für Psychosomatische Medizin und Psychotherapie (KPM), Universitätsmedizin Rostock, Germany ID: 242
/ Poster No. # 10: 045
Modalities: Audio, Multimodal Data, Time Series, Video, Other Methods: Other Application Domain: Health Modeling Neuro–Behavioral State Dynamics from Longitudinal Multimodal Recordings 1Deutsches Zentrum für Neurodegenerative Erkrankungen e. V. (DZNE), Germany; 2Helmholtz Munich, Neuherberg, Germany; 3University of Bonn, Medical Faculty, Bonn, Germany ID: 360
/ Poster No. # 10: 046
Modalities: Audio, Multimodal Data, Time Series, Video, Other Methods: Generative Models, Other Application Domain: Core Machine Learning, Health Consistent Representation Learning for Modeling Neuro-Behavioral State Dynamics 1Helmholtz Zentrum München Deutsches Forschungszentrum für Gesundheit und Umwelt ( GmbH ) Germany; 2German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany; 3University of Bonn, Medical Faculty, Bonn, Germany; 4Munich Center for Machine Learning ID: 288
/ Poster No. # 10: 047
Modalities: Image Methods: Other Application Domain: Matter Self-supervised Denoising Of Raw Tomography Detector Data For Improved Image Reconstruction Helmholtz-Zentrum Dresden - Rossendorf e. V., Germany ID: 222
/ Poster No. # 10: 048
Modalities: Image, Multimodal Data, Text Methods: Agentic AI Application Domain: Health Semantic Harmonization of Noisy Medical Imaging Cohorts via Vision-Language Model 1Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany; 2Imperial College London, UK; 3University of Zurich, Switzerland; 4ETH Zurich, Switzerland; 5Istanbul Medipol University, Turkey ID: 346
/ Poster No. # 10: 049
Modalities: Tabular Data Methods: Other Application Domain: Health Enhancing Clinical Immunology and Disease Diagnostics with Global Swarm Learning 1DZNE, Germany; 2Calico Life Sciences; 3Hewlett Packard Enterprise, Houston, Texas, USA; 4The Global Swarm Learning Group ID: 159
/ Poster No. # 10: 050
Modalities: Image, Video Methods: Generative Models Application Domain: Health Flow-aware Latent Diffusion for Synthetic Temporal Angiography 1LMU Munich, Germany; 2Heidelberg University, Germany ID: 306
/ Poster No. # 10: 051
Modalities: Graphs, Multimodal Data, Simulation Data, Text, Time Series Methods: Foundation Models, Graph Neural Networks Application Domain: Matter MatBind: Probing the multimodality of materials science with contrastive learning 1Institute for Advanced Simulations (IAS-9), Forschungszentrum Jülich GmbH, Germany; 2Helmholtz Institute for Polymer in Energy Applications Jena, Germany; 3Institute of Nanotechnology, Karlsruhe Institute of Technology, Germany; 4University of Cologne, Germany; 5RWTH Aachen University, Germany; 6Friedrich Schiller University Jena, Germany; 7Jena Center for Soft Matter, Germany; 8Center for Energy and Environmental Chemistry Jena, Germany; 9Jülich Supercomputing Centre, Forschungszentrum Jülich GmbH, Germany ID: 250
/ Poster No. # 10: 052
Modalities: Image Methods: Other Application Domain: Earth & Environment Creating a Benchmark Image Dataset for Reliable Quantification of Arctic Phytoplankton using AI 1Alfred-Wegener-Institut, Germany; 2University of Bremen, Germany ID: 235
/ Poster No. # 10: 053
Modalities: Graphs, Multimodal Data, Text Methods: Foundation Models Application Domain: Core Machine Learning, Health PromptGate: Federated Context Optimizationfor Dynamic VLM Gating in Open-Set MedicalImaging University of Bonn, University Hospital Bonn, Clinic for Diagnostic and Interventional Radiology, Bonn, Germany ID: 284
/ Poster No. # 10: 054
Modalities: Simulation Data, Time Series Methods: Physics-informed Machine Learning, Probabilistic Methods, Uncertainty Quantification Application Domain: Energy Neural Posterior Estimation for Empirical Power System Time Series Institute for Automation and Applied Informatics, Karlsruhe Institute of Technology, Germany ID: 398
/ Poster No. # 10: 055
Underconfident Predictions in MFVI Helmholtz Munich, Germany ID: 150
/ Poster No. # 10: 056
Modalities: Graphs, Simulation Data, Time Series Methods: Other Application Domain: Energy Differentiable Power-Flow Optimization 1Karlsruhe Institute of Technology, Germany; 2Helmholz AI, Karlsruhe ID: 220
/ Poster No. # 10: 057
Modalities: Simulation Data, Tabular Data, Other Methods: Generative Models, Probabilistic Methods Application Domain: Core Machine Learning, Health A generative model for dimensionality reduction with millions of features and few samples 1Helmholtz AI, Helmholtz Zentrum Munchen, Ingolstadter Landstraße 1,Neuherberg, Germany; 2Center for Health Data Science, Department of Public Health, University of Copenhagen, Copenhagen, Denmark; 3Department of Medical Sciences, University of Torino, Italy ID: 152
/ Poster No. # 10: 058
Modalities: Time Series Methods: Other Application Domain: Earth & Environment "Hello World": How to Have a Conversation with the Global Weather 1Karlsruhe Institute of Technology, Germany; 2Helmholtz AI ID: 335
/ Poster No. # 10: 059
Modalities: Image Methods: Other Application Domain: Health Segmentation of the intracranial cavity in T1-weighted MR images of the human head by a location-specific 3D U-Net 1Institute for Neuroscience and Medicine (INM-1), Forschungszentrum Jülich, Germany; 2C. and O. Vogt Institute for Brain Research, Medical Faculty, University Hospital, Heinrich-Heine-University of Düsseldorf, Germany; 3Institute for Anatomy I, Medical Faculty, University Hospital, Heinrich-Heine-University of Düsseldorf, Germany ID: 168
/ Poster No. # 10: 060
Modalities: Image Methods: Foundation Models Application Domain: Earth & Environment BioCUDA: An Accessible AI Pipeline for Fast, Accurate 3D MRI Reconstruction Across Species 1Data Science Group, Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, Bremerhaven; 2Integrative Ecophysiology, Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, Bremerhaven; 3Piraud Team, Helmholtz AI, Helmholtz Munich ID: 299
/ Poster No. # 10: 061
Modalities: Graphs, Simulation Data Methods: Foundation Models, Graph Neural Networks, Physics-informed Machine Learning Application Domain: Matter Machine-Learned Interatomic Potentials for Fe–Ni Alloys: Benchmarking MACE Models for Magnetism and Phase Stability 1Helmholtz-Zentrum Dresden-Rossendorf (HZDR); 2Center for Advanced Systems Understanding (CASUS) ID: 203
/ Poster No. # 10: 062
Modalities: Image, Multimodal Data, Tabular Data Methods: Other Application Domain: Health Hybrid Multimodal Late Fusion Frameworks for Robust bvFTD Classification in Imbalanced Dementia Cohorts 1German Center for Neurodegenerative Diseases (DZNE), Germany; 2Institute and Policlinic of Radiology, Pediatric Radiology and Neuroradiology, University Medical Center Rostock, Rostock, Germany; 3Department of Psychosomatic Medicine, Rostock University Medical Center, Rostock, Germany ID: 386
/ Poster No. # 10: 063
Amortising Inference and Meta-Learning Priors in Neural Networks 1Helmholtz AI, Germany; 2MCML, Germany; 3TUM, Germany; 4UTN, Germany ID: 337
/ Poster No. # 10: 064
Modalities: Image, Multimodal Data, Other Methods: Other Application Domain: Health SCHEMA: profiling Spatial Cancer HEterogeneity across modalities to benchmark Metastasis risk prediction 1Institute of Computational Biology, Helmholtz Munich, Germany; 2Institute of Lung Health and Immunity, Helmholtz Munich, Germany; 3University Hospital Tübingen, Germany; 4Karlsruhe Institute of Technology, Germany; 5TUM School of Life Sciences Weihenstephan,Technical University of Munich, Germany; 6TUM School of Computing, Information and Technology, Technical University of Munich, Munich, Germany ID: 282
/ Poster No. # 10: 065
Modalities: Graphs, Image, Simulation Data Methods: Physics-informed Machine Learning Application Domain: Information Machine Learning-Accelerated Prediction of Hydrogen Embrittlement fromTEM Microstructures 1Forschungszentrum Juüich GmbH; 2Helmholtz Zentrum Hereon, Germany ID: 133
/ Poster No. # 10: 066
Modalities: Time Series Methods: Foundation Models, Graph Neural Networks, Other Application Domain: Earth & Environment Multimodal Data-Driven Forecasting of Coastal Environmental Health in the Baltic Sea 1Christian-Albrechts-Universität zu Kiel (CAU), Germany; 2GEOMAR Helmholtz Centre for Ocean Research Kiel, Germany; 3Technische Universität Hamburg, Germany ID: 170
/ Poster No. # 10: 067
Modalities: Audio, Image Methods: Uncertainty Quantification, Other Application Domain: Earth & Environment Improvement of Whale Vocalization Detection with Temporal Context and Soft Labels 1Data Science Group, Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, Germany; 2Ocean Acoustics Group, Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research ID: 240
/ Poster No. # 10: 068
Modalities: Simulation Data Methods: Graph Neural Networks, Physics-informed Machine Learning Application Domain: Earth & Environment, Health Learning Coarse-Grained Potentials for Intrinsically Disordered Proteins Forschungszentrum Jülich, Germany ID: 135
/ Poster No. # 10: 069
Modalities: Text Methods: Other Application Domain: Information A modular Retrieval Augmented Generation (RAG) Toolbox for Building Transparent AI-Search Engines FZJ, Germany ID: 387
/ Poster No. # 10: 070
Better Uncertainties Don’t Guarantee Better Decisions 1HAI, Germany; 2TUM; 3relAI; 4MCML ID: 205
/ Poster No. # 10: 071
Modalities: Graphs Methods: Other Application Domain: Energy Attributing the effects of multiple line outages via Shapley values 1Institute of Climate and Energy Systems (ICE-1), Forschungszentrum Jülich, Germany; 2Amprion GmbH, Transmission System Operation, Germany ID: 234
/ Poster No. # 10: 072
Modalities: Image Methods: Probabilistic Methods, Uncertainty Quantification Application Domain: Aeronautics, Space & Transport, Earth & Environment Can Uncertainty Quantification Benefit From Label Embeddings? A Case Study on Local Climate Zone Classification 1German Aerospace Center, Germany; 2Interhyp; 3LMU Munich; 4TUM ID: 294
/ Poster No. # 10: 073
Modalities: Multimodal Data, Tabular Data Methods: Probabilistic Methods Application Domain: Health Benchmarking Taxonomic Classifiers: Bridging 16S rRNA metabarcoding and Shotgun Metagenomics for AI-Driven Microbiome Decoding 1Department of Tumor Immunology and Tumor Immunotherapy, Helmholtz Center for Translational Oncology (HI-TRON), Mainz, Germany; 2German Cancer Research Center (DKFZ), Heidelberg, Germany; 3Department of Hematology, Medical Oncology and Pneumology, University Medical Center, Mainz, Germany; 4Division of Microbiome and Cancer, German Cancer Research Center (DKFZ), Heidelberg, Germany; 5Epithelium Microbiome lnteractions, German Cancer Research Center (DKFZ), Heidelberg, Germany; 6Department of Systems Immunology, Weizmann Institute of Science, Rehovot, Israel ID: 289
/ Poster No. # 10: 074
Modalities: Graphs Methods: Generative Models, Probabilistic Methods Application Domain: Matter Toward Condition-Aware MOF Topology Prediction via Latent Generative Modeling 1Forschungszentrum Jülich GmbH, Germany; 2Helmholtz Institute for Polymers in Energy Applications Jena (HIPOLE Jena), Jena, Germany ID: 361
/ Poster No. # 10: 075
Modalities: Image Methods: Foundation Models Application Domain: Health Blur-aware Single-cell Segmentation and Feature Extraction for Drug-activity Measurement in Brightfield Organoid Imaging 1Helmholtz Zentrum München GmbH, Germany; 2Department for Translational Medical Oncology, NCT/UCC Dresden and DKFZ; 3Translational Medical Oncology, Carl Gustav Carus Faculty of Medicine, Dresden ID: 128
/ Poster No. # 10: 076
Modalities: Text Methods: Foundation Models, Generative Models Application Domain: Core Machine Learning, Health, Information Decoding Values: Measuring Culture Embedded in Large Language Models 1Perelyn GmbH, Germany; 2Technical University Munich, Germany ID: 274
/ Poster No. # 10: 077
Modalities: Image, Text Methods: Foundation Models, Generative Models Application Domain: Health Cytoarchitecture in Words: Weakly Supervised Vision–Language Modeling for Human Brain Microscopy 1Institute of Neuroscience and Medicine (INM-1), Research Centre Jülich, Germany; 2Helmholtz AI, Research Centre Jülich, Germany; 3Cécile & Oskar Vogt Institute for Brain Research, University Hospital Düsseldorf, Germamny; 4Computer Vision, Institute for Computational Visualistics, University of Koblenz, Germany ID: 366
/ Poster No. # 10: 078
Modalities: Multimodal Data, Text Methods: Agentic AI Application Domain: Core Machine Learning The (AI)^2 Consultant - A Co-consultant for AI Consultants Helmholtz Zentrum Munich, Germany ID: 312
/ Poster No. # 10: 079
Modalities: Image, Text Methods: Generative Models, Probabilistic Methods Application Domain: Health Accessible pipeline for open set dermatological captioning with scarce data 1Helmholtz AI, Helmholtz Zentrum München, Germany; 2Technical University of Munich, Klinikum rechts der Isar, Germany ID: 297
/ Poster No. # 10: 080
Modalities: Tabular Data, Time Series Methods: Other Application Domain: Earth & Environment Machine Learning for Rogue Wave Prediction in the Southern North Sea 1Helmholtz-Zentrum Hereon, Germany; 2Federal Waterways Engineering and Research Institute (BAW); 3Helmholtz AI ID: 279
/ Poster No. # 10: 081
Modalities: Time Series Methods: Other Application Domain: Health Nanopore- and AI-Empowered Microbial Viability Inference Helmholtz Munich, Germany ID: 301
/ Poster No. # 10: 082
Modalities: Image, Multimodal Data, Time Series Methods: Other Application Domain: Aeronautics, Space & Transport, Earth & Environment Counterfactual Ship Attribution in Cloud Fields Using Deep Learning and Satellite Observations German Aerospace Center (DLR), Germany ID: 208
/ Poster No. # 10: 083
Modalities: Time Series Methods: Other Application Domain: Energy, Earth & Environment RenewBench: Real Energy Data You Can Actually Use 1Karlsruhe Institute of Technology (KIT), Germany; 2Helmholtz-Center Hereon, Germany; 3Helmholtz AI, Germany ID: 315
/ Poster No. # 10: 084
Modalities: Text Methods: Agentic AI, Generative Models Application Domain: Health, Information An AI System for Robust Plain Language Translation of Clinical Trials 1Institute of Computational Biology, Computational Health Center, Helmholtz Munich, Germany; 2German Center for Diabetes Research, Germany ID: 309
/ Poster No. # 10: 085
Modalities: Image, Text Methods: Foundation Models Application Domain: Health K-MaT: Knowledge-Anchored Manifold Transport For Cross-Modal Prompt Learning In Medical Imaging University of Bonn, University Hospital Bonn, Clinic for Diagnostic and Interventional Radiology, 53127 Bonn, Germany ID: 154
/ Poster No. # 10: 086
Modalities: Multimodal Data, Simulation Data, Time Series Methods: Other Application Domain: Health Analysis of immune responses in renal cell carcinoma with synergistic ex vivo and in silico simulation tumor models 1German Cancer Research Center (DKFZ), Helmholtz Institute for Translational Oncology (HI-TRON), Tumor Immunology and Tumor Immunotherapy, Mainz, Germany; 2University Medical Center Mainz, Department of Hematology and Medical Oncology, Mainz, Germany; 3Faculty of Biosciences, Ruprecht Karls University, Heidelberg, Germany; 4National Center for Tumor Diseases, Department of Medical Oncology, Heidelberg, Germany; 5University Medical Center Mainz, Institute for Pathology, Mainz, Germany; 6University Medical Center Mainz, University Cancer Center (UCT), Mainz, Germany ID: 120
/ Poster No. # 10: 087
Modalities: Time Series Methods: Probabilistic Methods, Uncertainty Quantification Application Domain: Energy Bad Data Detection and Signal Recovery in Large-Scale ICT-Platforms 1ICE-1, Institute of Climate and Energy Systems, Forschungszentrum Jülich, 52428 Jülich, Germany; 2RWTH Aachen University, Aachen 52056, Germany; 3JARA-Energy, Jülich 52425, Germany ID: 119
/ Poster No. # 10: 088
Modalities: Time Series Methods: Physics-informed Machine Learning, Probabilistic Methods Application Domain: Energy System Characterization by Affine Operators in Physics-Informed Gaussian Processes 1ICE-1, Institute of Climate and Energy Systems, Forschungszentrum Jülich, 52428 Jülich, Germany; 2RWTH Aachen University, Aachen 52056, Germany; 3JARA-Energy, Jülich 52425, Germany ID: 223
/ Poster No. # 10: 089
Modalities: Simulation Data Methods: Physics-informed Machine Learning, Other Application Domain: Information, Matter NEOPIC: A Neural Operator Framework for Particle-based Kinetic Plasma Simulations 1Forschungszentrum Jülich GmbH, Germany; 2Helmholtz Zentrum Dresden-Rossendorf, Germany ID: 384
/ Poster No. # 10: 090
Learning PDE Dynamics from Small Data: Out-of-Distribution Generalization of Deep Learning Surrogates 1Institute for Advanced Simulations -- Materials Data Science and Informatics (IAS-9), Forschungszentrum Juelich GmbH, Juelich 52425, Germany; 2Chair of Materials Data Science and Materials Informatics, Faculty 5 -- Georesources and Materials Engineering, RWTH Aachen University, Aachen 52056, Germany ID: 163
/ Poster No. # 10: 091
Modalities: Graphs, Image, Multimodal Data, Simulation Data, Tabular Data, Time Series, Other Methods: Foundation Models, Physics-informed Machine Learning, Probabilistic Methods, Other Application Domain: Energy, Information, Matter Human-Explainable, Compact, Clustering-based Latents for Fast Proton Energy Spectra Estimation 1Helmholtz-Zentrum Dresden-Rossendorf e.V., Germany; 2Center for Advance Systems Understanding (CASUS), Görlitz, Saxony, Germany; 3TUD Dresden University of Technology, 01062 Dresden, Germany; 4Technische Universität Chemnitz, Institute of Physics, Chemnitz, Saxony, Germany ID: 263
/ Poster No. # 10: 092
Modalities: Audio, Image, Multimodal Data, Tabular Data, Text Methods: Generative Models, Other Application Domain: Core Machine Learning, Information Collaborative Research Data Management for AI Hof University of Applied Sciences, Germany ID: 280
/ Poster No. # 10: 093
Modalities: Multimodal Data Methods: Other Application Domain: Health From FAIR Multimodal Data Infrastructure to AI-Driven Clinical Interpretation: The MINDset Platform for Precision Psychiatry 1Institute of Neuroscience and Medicine 4 (INM-4), Forschungszentrum Jülich GmbH, Jülich, Germany; 2Department of Psychiatry, Psychotherapy and Psychosomatics, University Hospital Aachen,RWTH Aachen University, Aachen, Germany; 3Department of Computer and Information Science and Engineering, University of Florida, USA; 4Faculty of Medical Engineering and Technomathematics, FH Aachen University Applied Sciences, Aachen, Germany; 5Department of Neurology, RWTH Aachen University, Aachen, Germany; 6Department of Psychiatry II, Ulm University and BKH Günzburg, Germany. ID: 238
/ Poster No. # 10: 095
Modalities: Graphs, Time Series Methods: Generative Models, Graph Neural Networks Application Domain: Aeronautics, Space & Transport Measuring Diversity in Multi-Agent Motion Generation 1Karlsruhe Institute of Technology, Germany; 2Aumovio SE ID: 140
/ Poster No. # 10: 096
Modalities: Graphs, Image, Simulation Data, Time Series, Video Methods: Generative Models, Graph Neural Networks, Physics-informed Machine Learning Application Domain: Earth & Environment GEOMAGFOR - Geomagnetic Core Field Forecasting: Excursions and Polarity Reversals 1Institute of Fluid Dynamics, Helmholtz-Zentrum Dresden-Rossendorf (HZDR), Germany; 2Geomagnetism, GFZ Helmholtz Centre for Geosciences, Potsdam, Germany ID: 226
/ Poster No. # 10: 097
Modalities: Simulation Data Methods: Probabilistic Methods, Uncertainty Quantification Application Domain: Health Clinical Evidence to Individualized Care: AI Powered Clinical Decision Support Based on Predicted Individual Treatment Effect (PITE) 1Faculty for Informatics and Data Science, Regensburg University, Germany; 2MRC Biostatistics Unit, University of Cambridge, UK ID: 275
/ Poster No. # 10: 098
Modalities: Other Methods: Generative Models Application Domain: Earth & Environment, Information Integrating an LLM Chatbot into an Existing Web Frontend via DASF and MCP Helmholtz-Zentrum Hereon, Germany ID: 381
/ Poster No. # 10: 099
CapTrack: Multifaceted Evaluation of Forgetting in LLM Post-Training 1Thomson Reuters Foundational Research; 2Tübingen AI Center, University of Tübingen; 3Helmholtz Munich; 4MCML, Technical University of Munich; 5Imperial College London ID: 214
/ Poster No. # 10: 100
Modalities: Time Series Methods: Other Application Domain: Matter Patch-MLP-Based Predictive Control: Simulation of Upstream Pointing Stabilization for PHELIX Laser System 1Helmholtz-Zentrum Dresden-Rossendorf, Dresden, Germany; 2GSI Helmholtzzentrum für Schwerionenforschung, Darmstadt, Germany; 3Amplitude laser group—Dresden operations, Dresden, Germany; 4Extreme Light Infrastructure—Nuclear Physics, National Institute for Physics and Nuclear Engineering, Ilfov, Romania; 5University of Bucharest, Ilfov, Romania; 6Engineering and Applications of Lasers and Accelerators Doctoral School (SDIALA), National University of Science and Technology Politehnica of Bucharest, Bucharest RO-060042, Romania; 7Technische Universität Dresden, Dresden, Germany; 8Chemnitz University of Technology, Chemnitz, Germany ID: 1398
/ Poster No. # 10: 101
Modalities: Graphs Methods: Foundation Models, Graph Neural Networks Application Domain: Health Evaluating 3D Molecular Representation Learning with Multipolar Atom Types 1University of Warsaw; 2Helmholtz Munich HELMHOLTZ AI PROJECT CALL AWARDEES 2025
ID: 1403 / Poster No. # 10: 102 Modalities: Image, Multimodal Data Methods: Foundation Models, Other Application Domain: Core Machine Learning, Health, Information, Matter LIVR: Learnable Implicit Volumetric Representations for High-resolution 3D Images 1Division of Medical Image Computing, German Cancer Research Center (DKFZ), Heidelberg, Germany; 2Institute of Materials Physics, Helmholtz-Zentrum Hereon, Geesthacht, Germany; 3Institute of Metallic Biomaterials, Helmholtz-Zentrum Hereon, Geesthacht, Germany HELMHOLTZ AI PROJECT CALL AWARDEES 2025
ID: 1404 / Poster No. # 10: 103 Modalities: Multimodal Data, Simulation Data, Tabular Data, Time Series Methods: Physics-informed Machine Learning, Probabilistic Methods, Uncertainty Quantification Application Domain: Earth & Environment The AEON-UP Helmholtz AI Project - Adaptive Environmental Prediction System using Neural Processes for Urban Air Quality Prediction 1Helmholtz-Zentrum Hereon, Germany; 2RIFS Forschungsinstitut für Nachhaltigkeit, Germany HELMHOLTZ AI PROJECT CALL AWARDEES 2025
ID: 1411 / Poster No. # 10: 104 Modalities: Image Methods: Foundation Models Application Domain: Health SCALE.FM: Multi-Scale Foundation Models for Precision Neuroimaging 1Deutsches Zentrum für Neurodegenerative Erkrankungen (DZNE), Bonn, Germany; 2Helmholtz Zentrum München - Deutsches Forschungszentrum für Gesundheit und Umwelt (HMGU), Munich, Germany HELMHOLTZ AI PROJECT CALL AWARDEES 2025
ID: 1416 / Poster No. # 10: 105 Modalities: Simulation Data, Time Series Methods: Foundation Models, Reinforcement Learning, Probabilistic Methods, Uncertainty Quantification Application Domain: Matter Machine Learning Green Functions for Magnetic Materials and Spintronics (MLGREEN) 1Helmholtz-Zentrum Dresden-Rossendorf, Germany; 2Forschungszentrum Jülich, Germany HELMHOLTZ AI PROJECT CALL AWARDEES 2025
ID: 1418 / Poster No. # 10: 106 Modalities: Graphs, Multimodal Data, Simulation Data, Text Methods: Generative Models, Physics-informed Machine Learning Application Domain: Health, Matter Generative AI Framework for Steered Protein Structure Prediction via NMR Chemical Shift Inputs 1Helmholtz Munich, Germany; 2Technical University of Munich, Germany ID: 1149
/ Poster No. # 10: 107
Modalities: Image Methods: Other Application Domain: Health A decentralized Swarm Learning framework for 90-Day outcome prediction for acute ischaemic stroke 1DZNE, Germany; 2CISPA, Germany | ||
