Institute for Comparative Media and Communication Studies (CMC)
Contact: Charlotte Spencer-Smith
Email: Charlotte.Spencer-Smith(at)oeaw.ac.at
Research: The research group “Media Accountability and Media Change” analyses technological innovations and structural changes in global, European and national media systems. This includes the role of AI and algorithmic decision-making in content curation and public communication, as well as their societal implications. The group is interested in communication ethics in generative AI, including questions of governance, public interest and media diversity. Current research focusses on the alignment and moderation practices of popular consumer LLMs and their effect on discourse and public spheres. Of particular interest is how rules and guardrails influence how models behave and 'speak'.
Data Keywords: text
Task Keywords: large language models (LLMs), generative AI, techniques for identifying alignment
Institute of Technology Assessment (ITA)
Contact: Doris Allhutter
Email: doris.allhutter(at)oeaw.ac.at
Research: Scientific technology assessment researches the impacts of new technologies on society, the environment and the economy and advices policy-makers, administration and the public on issues of technology policy. ITA studies practices of machine learning and the development, application and governance of artificial intelligence and algorithms in a variety of realms. Amongst them are the public sector, education and research, work, business and society at large. We explore the potentials and challenges of machine learning and AI technologies with regard to ethical questions, including technological dependencies, lack of transparency and verifiability. Our research focuses on the intersection between ML/AI and the social values of equality, privacy, democracy, sustainability, human autonomy, and security.
Data Keywords: text, image, multimodal, open (government) data
Task Keywords: human-centered assessment, normativity of ML practices, governance of algorithmic systems and AI, AI literacy, generative AI, classification, prediction, semantic embeddings
The Space Research Institute (IWF)
Contact: Amit Reza, Academy Scientist in Machine Learning
Email: amit.reza(at)oeaw.ac.at
Research: The Space Research Institute (IWF) has recently adopted machine learning (ML), applying it across core research areas, including exoplanet weather and climate, exoplanet characterization and evolution, proto-planetary disks, astrochemistry, and satellite laser ranging. IWF has developed an ML emulator that produces complete radiative transfer models in real-time, helping to resolve the degeneracy problem in deriving protoplanetary disk properties from spectral energy distributions. The rapid generation of these models enables full Bayesian inference of disk parameters, revealing uncertainties and degeneracies that traditional methods often overlook. In exoplanet atmospheric modeling, IWF has introduced a novel DeepSets-based ML method that integrates chemical opacity data into general circulation models. This approach allows for more efficient and accurate radiative transfer calculations in atmospheres with non-equilibrium chemistry, representing a significant step forward for realistic climate simulations of exoplanets such as hot Jupiters. To advance understanding of planetary evolution, IWF has developed an ML-based regression tool that estimates atmospheric mass-loss rates and improves interpolation accuracy across large grids of hydrodynamic models compared to classical methods. Finally, IWF’s ML-based classifiers for analyzing light-curve data from satellite laser ranging support improved space debris classification and monitoring.
Data Keywords: time series, spectral data, spatio/temporal
Task keywords: regression, classification, dimensionality reduction, simulation-based inference, anomaly detection
Institute for Habsburg and Balkan Studies (IHB)
Contact: Doris Gruber
Email: doris.gruber(at)oeaw.ac.at
Research: The IHB is a leading non-university institution for the historical research of the Habsburg Monarchy, its successor states and the Balkan region. With AI-supported methods, we have been exploring new paths in the identification, extraction and categorization of early modern images, texts and their relationships, especially by focusing on travelogues about the Ottoman Empire as part of the projects Travelogues and ONiT. Many projects at the institute create full texts of handwritten and printed material using HTR (Transkribus). Further applications of AI for the collection, preservation and analysis of cultural heritage are being explored.
Data keywords: text, image, multimodal
Task keywords: transformer models, transfer learning, contrastive learning, embedding-based retrieval, Handwritten Text Recognition (HTR), Transkribus
Vienna Institute of Demography (VID)
Contact: Bernhard Rengs & Isabel Gerstner
Email: bernhard.rengs(at)oeaw.ac.at & isabel.gerstner(at)oeaw.ac.at
Research: At the Vienna Institute of Demography, we focus on understanding demographic changes, including population ageing, migration patterns, fertility trends, and labour market dynamics. As AI and Machine Learning are slowly but increasingly explored in demography, we aim to assess their potential for advancing predictive modelling, policy simulations, and trend analysis. We are particularly interested in how machine learning methods can complement classical demographic approaches, especially in forecasting population dynamics, analysing socio-economic stratification, and modelling spatio-temporal demographic trends. While data availability and methodological robustness remain key considerations, we strive to be at the forefront of evaluating AI-driven innovations in demographic research.
Data Keywords: tabular, time series, spatio-temporal, networks, survey data, register data, microdata, longitudinal data
Task Keywords: Bayesian statistics, time series forecasting, regression, classification
Institute For Interdisciplinary Mountain Research (IGF)
Contact: Mathieu Gravey (Junior group leader - Digital Landscape)
Email: mathieu.gravey(at)oeaw.ac.at
Research: At IGF, my research group focuses on multiple remote sensing and geostatistics applications. We utilize existing models and develop custom solutions to meet our specific needs. For instance, we employ transformer-based approaches to explore the spatial distribution of complex underground structures. This allows us to assess water distribution, connectivity for optimal water supply, and contamination propagation. Our multimodal methods extend to disease management and mineral resource estimation applications. Additionally, we apply machine learning techniques for remote sensing data to classify clouds, snow, vegetation, and other environmental features and explore links between them.
Data Keywords: tabular, non-grided spatial data, images, rasters, timeseries
Task Keywords: generative, classification, regression, information retrieval, stochastic simulation, estimation, uncertainty quantification
Institute for Medieval Research (IMAFO)
Contact: Jan Odstrčilík
Email: jan.odstrcilik(at)oeaw.ac.at
Research: The Institute for Medieval Research focuses on the period between c. 300 and c. 1500 in Europe and the Mediterranean World as far as the Euphrates. It combines the documentation and editing of medieval sources with a source-based approach to fundamental research questions. One of its main research areas is the formation formation of medieval identities. The current topics include the emergence of European peoples during the transformation of the Roman World, the cultural profile of Carolingian Europe and its transformation, the changing role of ethnicity, the impact of Christianity on the formation of particular identities in a comparative perspective, and the significance of translations into vernacular languages in late medieval Central Europe.
Currently, Machine Learning is applied in two main areas: Handwritten Text Recognition of medieval documents in multiple languages and Archaeology, specifically the classification of glass beads. Additionally, further areas of machine learning are being explored, particularly in regard to the automatic detection of text reuse.
Data Keywords: images (Manuscripts, Glass Beads), text, tabular data
Task Keywords: Handwritten Text Recognition (HTR), PyLaia, Kraken, Transkribus, eScriptorium, classification, convolutional neural networks, recursive neural networks, semantic embeddings
Research Center for Molecular Medicine (CeMM)
Contact: Christoph Bock (Principal Investigator)
Email: cbock(at)cemm.oeaw.ac.at
Research: ML/AI research at CeMM focuses on understanding the biological functions that underlie organ function and disfunction in health and disease. Typical datasets include single-cell gene expression profiles of tissues and tumors, microscopic imaging data of organs, and perturbational datasets (e.g., CRISPR and drug screens). Methodological work focuses on multimodal models, times series, perturbations and causality, and multiscale networks. The overarching goal is to advance molecular medicine for diseases such as immune disorders and infections, cancer and metabolic disorders (https://www.cemm.at/about/mission).
Data Keywords: tabular data, graphs, images, spatial data, time series, text, multimodal
Task Keywords: regression, classification, generative, reinforcement, clustering, dimensionality reduction, anomaly detection
Erich Schmid Institute of Materials Science (ESI)
Contact: Claus O. W. Trost (Post-Doc)
Email: claus.trost(at)oeaw.ac.at
Research: Materials can be produced through various routes and combinations, resulting in diverse mechanical and functional properties (e.g., magnetic, self-healing). Data-centric approaches are essential to allow for predictions beyond theoretical models, experimental intuition and simulations. Digital lab notebooks play a key role in generating FAIR data, allowing machine learning to extract insights even from otherwise considered "failed" experiments.
We aim to enable more targeted experiments by enhancing materials testing methods. This can be achieved by predicting difficult-to-measure (or costly) parameters or by applying classification and clustering methods to analyse high-throughput experimental data, thereby disentangling local material properties. To achieve this, multi-fidelity data from simulations and experiments are utilised.
By leveraging game theory-based methods to explain model predictions, we enhance trust in our models, guiding further targeted experiments and simulations for deeper physical understanding.
Data Keywords: multi-fidelity tabular data, images, small data sets, imbalanced data
Task keywords: regression, classification, clustering, anomaly detection, explainable machine learning
Acoustics Research Institute (ARI)
Contact: Nicki Holighaus (Speaker - Machine Learning in Acoustics)
Email: nicki.holighaus(at)oeaw.ac.at
Research: We study the theoretical foundations and applications of artificial intelligence in acoustics through interdisciplinary collaboration. Our AI methods enhance researchers’ understanding of animal communication, improve personalized spatial audio experiences through headphones, and facilitate the automated processing and generation of speech and music signals. Furthermore, we employ mathematical methods to analyze information flow in artificial neural networks, offering new insights for future AI developments..
Data Keywords: time series, images, video, meshes
Task Keywords: signal enhancement, classification, parameter estimation, generative
Marietta Blau Institute for Particle Physics (MBI Vienna)
(formerly Institute of High Energy Physics (HEPHY))
Contact: Claudius Krause (Group Leader - Machine Learning in Particle Physics)
Email: claudius.krause(at)oeaw.ac.at
Research: In particle physics, we are interested in better understanding nature's fundamental building blocks and interactions by reproducing the early universe in high-energy collider experiments like the LHC at CERN or Belle2 at KEK in Japan. The vital link between theoretical understanding and experimental measurements is based on a sophisticated simulation and analysis chain which is increasingly supported by modern machine learning (ML) techniques. On the experimental side, ML is used for tasks such as reconstructing particles from raw detector hits, selecting interesting events from a vast number of uninteresting ones occurring in the same collision, and distinguishing particles originating from different underlying processes. On the theoretical side, ML helps enhance the quality and efficiency of simulations, identify statistically optimal observables, and develop new methods for anomaly detection—essentially finding the needle in the haystack without prior knowledge of what either looks like. We are also working on the ML-assisted analysis of gravitational wave data from experiments like VIRGO or the upcoming Einstein Telescope.
Data Keywords: tabular data, point clouds, graphs, images, time series
Task Keywords: generative, classification, regression, anomaly detection, dimensionality reduction, time series forecasting, simulation-based inference
Austrian Centre for Digital Humanities (ACDH)
Contact: Elisabeth Eder
Email: elisabeth.eder(at)oeaw.ac.at
Research: At the Austrian Centre for Digital Humanities, we work at the intersection of technology, linguistics, literary studies, prosopography and musicology. The institute fosters the adoption of AI methods in humanities research. We apply machine learning (ML) methods on objects of music, texts and language for various tasks such as natural language processing, automated text and layout recognition, information extraction, data generation and processing.
Examples of projects include investigating language usage and language change by exploring and extending the WBÖ (Wörterbuch der historischen bairischen Mundarten in Österreich und Südtirol), intra- and extratextual relationships by exploring prosopographic datasets as well as numerous digital editions of authors connected to Austria.
With its wide-ranging scholarly and technical expertise, its high-quality, manually curated data material, and its numerous collaborations, the ACDH is ideally positioned to contribute to the application, evaluation and further development of domain-specific AI models and AI-based workflows.
Data Keywords: text, images, historical data, prosopographical data, linguistic data
Task Keywords: knowledge extraction, NLP, automatic text/layout recognition, text generation
Johann Radon Institute of Computational and Applied Mathematics (RICAM)
Contact: Dr. Thomas Dittrich
Email: Thomas.Dittrich(at)oeaw.ac.at
Research:
Machine learning research at the Johann Radon Institute of Computational and Applied Mathematics (RICAM) focuses on developing new algorithms as well as studying the theoretical aspects of machine learning. We operate at the interface of rigorous mathematical analysis, modern machine learning, and scientific applications. Our research spans approximation theory, sampling complexity, inverse problems, and algorithmic developments for scientific computing, optimal control, and quantum many-body systems.
A central theme is a mathematical understanding of modern deep learning algorithms, addressing reliability, approximation power, sampling efficiency, stability, and fundamental limitations and strengths of neural network methods in numerical analysis and physics.
Data Keywords: high-dimensional data, infinite-dimensional data
Task Keywords: scientific machine learning, operator learning, machine learning theory