2025 software,ia

Developing AI Models on the EOS Cloud

AI4EOSC

Summary

The European Open Science Cloud (EOSC) environment aims to offer the European scientific community a set of cloud services to enhance their experience working with data in research, innovation and education.

In recent years, in particular from the second decade of the 21st century onwards, there has been an unprecedented increase in the interest and use of Artificial Intelligence (AI), machine learning (ML) and deep learning (DL) for their amazing results. In this context, comes the Artificial Intelligence for the European Open Science Cloud(AI4EOSC) project, which focuses on the creation of innovative services for the development of models and applications of these techniques, integrated in a single platform in the EOSC environment. These features include federated, distributed and split learning, traceability management of the elements involved or the use of a serverless architecture.

AI4EOSC started on 1 September 2022 and will last for 3 years with a budget of approximately 5 million euros. This work will be carried out by a consortium of 10 institutions from 6 different countries, of which the Advanced Computing group of the Physics Institute of Cantabria (IFCA) will be coordinating the efforts.

This project has received funding from the European Union’s Horizon Research and Innovation programme under Grant agreement No. 101058593.

Main Features

  • European Open Science Cloud
  • Secure and efficient distributed computing for Artificial Intelligence
  • Services for distributed use of sensitive data
  • Full machine learning experiment traceability
  • Model training and inference in the cloud
AI4EOSC CONTACT

    Our contribution

    Predictia's contribution to this project is focused on the following activities:

    • Co-design of the architecture: establishing functional and non-functional requirements based on the use cases to jointly design the AI platform and its components, which in turn integrate the proposed advanced features.
    • Traceability and metadata: defining a standard model for the management of the domain metadata generated in each experiment and the traceability of all the specific elements such as datasets, hyperparameters or development frameworks.
    • Use cases: development and integration of use cases with AI models using the services provided by AI4EOSC. In particular, we will design and implement state-of-the-art DL architectures for remote sensing or nowcasting.
    • Dissemination and promotion: give visibility to the services developed during the project, look for synergies with other initiatives or projects where AI4EOSC can have a positive impact.

    Browse our projects

    We use cookies to enhance the user experience. By clicking any link on this page, you are giving your consent.