Funding for research with Azure services

/

• Marina Dütsch | FLEXWEB

  Organisational unit: Department of Meteorology and Geophysics

  Abstract (in German):

  Flexpart (FLEXible PARTicle dispersion model) ist ein numerisches Modell, das die Ausbreitung von Gasen und Aerosolen in der Atmosphäre simuliert. Das Ziel dieses Projekts war, ein Web Service (FLEXWEB) entwickeln, bei dem Flexpart über eine Webseite laufen gelassen werden kann. Flexpart sollte mit Hilfe eines Kubernetes Clusters Trajektorien berechnen und den Usern diese Ergebnisse leicht zugänglich machen. Ein wichtiger Schritt dabei war die Containerisierung der Arbeitsschritte und diese mit den relativen großen Eingangsdaten abzustimmen. Eine erste Version von FLEXWEB haben wir erfolgreich zum Laufen gebracht, allerdings ist es uns nicht gelungen, das Service skalierbar zu machen, so dass mehrere User es gleichzeitig nutzen können

  Flexpart Entwicklung an der Universität Wien

  Category: Geförderte Forschungsprojekte 2024
• Wolfgang Klas | FactCheck

  Organisational unit: Multimedia Information Systems research group, Faculty of Computer Science

  Abstract:

  FactCheck is an ongoing research project conducted by the Research Group on Multimedia Information Systems within the Faculty of Computer Science. The project aims to identify conflicts in data available on the web. These conflicts may exist in various formats, such as textual (e.g., paragraphs in HTML documents) or multimedia (e.g., news segments in videos). To extract this conflict information, we will employ a combination of Semantic Web approaches (e.g., structured data) and state-of-the-art artificial intelligence technologies (e.g., named entity recognition and entity linking). The comparison processes for this information will involve both human intelligence and feedback, which is why we will also investigate methods for user identity and user management, such as Azure Entra ID. We are considering a hybrid approach to deploying the FactCheck prototypes. This will involve using both scalable Azure services (e.g., cognitive services like AI Video Indexer and user management solutions) and the existing on-premises infrastructure (e.g., virtual machines or databases) at the University of Vienna. This approach aims to achieve an appropriate balance between security, privacy, and cost. Some components may be containerized to ensure that the deployment remains highly flexible and modular, facilitating easier deployment on both Azure and local infrastructure.

  Category: Geförderte Forschungsprojekte 2024
• Oliver Wieder | Revolutionizing Olfactory Perception Mapping: A Contrastive Learning Graph Neural Network Approach

  Organisational unit: Department of Pharmaceutical Sciences

  Abstract:

  Abstract will be added soon

  Category: Geförderte Forschungsprojekte 2024
• Abert Claas | Very Largescale Distributed Micromagnetic Research Tools

  Organisational unit: Physics of Functional Materials

  Abstract (in German):

  Im Rahmen des Projekts wurde untersucht, wie gut sich die Microsoft Azure Cloud für wissenschaftliche Berechnungen im Bereich der Mikromagnetik eignet. Dafür wurden verschiedene virtuelle Maschinen getestet, sowohl mit Prozessoren als auch mit Grafikprozessoren. Ein besonderer Fokus lag auf der Nutzung von kostengünstigen Spot Instanzen, die jedoch gelegentlich unterbrochen werden können. Um die Arbeit mit den Simulationen zu vereinfachen, wurden hilfreiche Tools entwickelt, die Abläufe automatisieren und so Zeit sparen. Während des Projekts konnten außerdem mikromagnetische Simulationen zur Optimierung von magnonischen Geräten erfolgreich durchgeführt werden, die wertvolle Erkenntnisse lieferten. Anstelle ursprünglich geplanter Studien zur Nutzung mehrerer Grafikprozessoren lag der Schwerpunkt darauf, die Cloud-Lösungen umfassend zu bewerten. Zusätzlich wurde eine Bachelorarbeit verfasst, die zur Entwicklung der Automatisierungstools beitrug. Das Projektbudget von 2.000 EUR wurde fast vollständig genutzt, und die Ergebnisse zeigten, dass die Microsoft Azure Infrastruktur ein großes Potenzial für wissenschaftliche Anwendungen bietet.

  Category: Geförderte Forschungsprojekte 2024
• Xin Huang | selscape: Automated and Distributed Pipelines for Investigating the Landscape of Natural Selection from Large-scale Genomic Datasets

  Organisational unit: Department of Evolutionary Anthropology

  Abstract:

  This project developed three Snakemake pipelines for detecting balancing selection, positive selection, and inferring the distribution of fitness effects. Azure Batch was tested for cloud deployment, and the first pipeline was successfully implemented in the cloud. The remaining pipelines are ready for deployment using insights gained from the first pipeline’s testing. Key results include contributions to three studies, showcasing the pipelines' effectiveness in analyzing genomes and exploring genetic diversity. Despite challenges with inadequate documentation for integrating Snakemake with Azure Batch, the project goals were partially achieved, with development carried out conservatively on local servers due to the novelty of cloud integration. Future work will focus on fully deploying all pipelines in the cloud and expanding their applications for large-scale genomic analyses.

  Category: Geförderte Forschungsprojekte 2024
• Dylan Paltra | MULTIREP – Multidimensional Representation: Enabling An Alternative Research Agenda on the Citizen-Politician Relationship

  Organisational unit: Department of Government

  Abstract:

  The “MULTIREP” project aims to enable an alternative approach to studying the citizen-politician relationship. It focuses primarily on how citizens conceptualize representation. A mixed-methods approach combines qualitative methods (focus groups and one-to-one interviews with citizens) and quantitative methods in five countries (ca. 2.000 respondents in each), focusing on natural language processing approaches. In a multinational and multilingual mass survey in five countries, including 10.000 participants, we want to improve the current survey methodology by analyzing respondents’ answers to open-ended questions using different machine-learning approaches. During the funding period, the project team was able to conduct the survey, collecting rich text data from representative samples of the public. The team used the Azure infrastructure to analyze the open-ended text answers preliminary by prompting large-language models. These results complement a theoretically induced coding scheme, which will be used later in the analysis. Besides the already established dimensions of representation, the team found that citizens conceptualize representation very much in formalistic terms. The team plans to continue the usage of Microsoft Azure to thoroughly analyse the open-ended text answers, making use of not only large-language models but also more established natural language processing approaches.

  Category: Geförderte Forschungsprojekte 2024
• Miguel Angel Rios Gaona | Controlled Machine Translation with Large Language Models for the Technical Domain

  Organisational unit: Centre for Translation Studies

  Abstract:

  Large Language Models (LLMs) have shown promising results on machine translation for high resource language pairs and domains. However, in specialised domains (e.g. medical) LLMs have shown lower performance compared to standard neural machine translation models. The consistency in the machine translation of terminology is crucial for users, researchers, and translators in specialised domains. In this study, we compare the performance between baseline LLMs and instruction-tuned LLMs in the medical domain. In addition, we introduce terminology from specialised medical dictionaries into the instruction formatted datasets for fine-tuning LLMs. The instruction-tuned LLMs significantly outperform the baseline models with automatic metrics, and quality estimation. Moreover, the instruction-tuned LLMs produce fewer errors compared to the baseline based on automatic error annotation.

  Category: Geförderte Forschungsprojekte 2024