October 11, 2024

The AI4EOSC project presented its significant advancements in the processing of research data with artificial intelligence and machine learning during the EGI conference 2024, held in the Italian city of Lecce from September 30 to October 4.

The project coordinator Álvaro López García presented AI4EOSC during a session that also featured updates on topics such as MLOps (by Valentin Kozlov), evaluation of open source LLMs with MLflow (by Lisana Berberi), a comprehensive comparison of federated learning (FL) frameworks (by Khadijeh Alibabaei) and the implementation of the AI4EOSC distributed platform (by Saúl Fernández).

AI4EOSC was represented in this important event with numerous contributions, including a booth dedicated to the project, along with numerous talks, four demos, and a poster.

Regarding the demonstrations, Fernando Aguilar showcased a tool for evaluating the FAIR principles, called FAIR EVA. Judith Sáinz-Pardo presented the implementation of secure personalized federated learning within the AI4EOSC platform, demonstrating two use cases: one related to the project (agrometeorology) and another on medical data. Furthermore, Amanda Calatrava presented the advancements in AI4Compose to visually compose inference pipelines that execute on OSCAR clusters. In addition, Borja Esteban Sanchís focused his demonstration on the development of templates for reproducible science.

The project was also presented with its own dedicated booth, in which different videos and tutorials on the use of the platform were showcased, as well as samples of its functionality.

Finally, the poster presented by Marcin Plociennik on the use of the AI4EOSC platform for integrated plant protection won the award for the best poster of the conference, and the team won the contest for the best photo.