Data Week 2022 was held both online (May 25th to June 3rd) and physically (June 8th – 9th) this year. Data Week is the spring gathering of the European Big Data and Data Driven AI research and innovation communities, where the participants share knowledge and results, discuss topics of common interest, find synergies, build new collaborations and identify new challenges and recommendations. Data Week is co-organized by Big Data Value Association and the EUHubs4Data project. This year’s main theme was “Towards an innovative, trusted, and fair European data economy”.

On Wednesday, May 25th, MARVEL’s Project Coordinator Prof. Sotiris Ioannidis e-welcomed Dr. Manfredo Atzori, Scientific Coordinator of ExaMode, Dr. Nikolaos Passalis, Technical Manager of OpenDR, and Dr. Paulo Figueiras, senior Researcher of VesselAI at a workshop where they explored “The challenges of the extreme-scale multi-modal analytics applications” in the context of the research projects they participate. The session explored the latest EU research outcomes related to Extreme Data Analytics applications in smart cities, healthcare, robotics, and the maritime industry. The participants focused on the key challenges addressed in each sector in an effort to showcase the lessons they learned the hard way and encourage other researchers to take them into consideration.
In particular, the key outcomes of the session were:

• • Pay attention to ethics and privacy concerns and account for a dedicated task on this topic which is a matter of high concern for all users.
  • Integration is always complex. Do not try to do everything in parallel, rather pace your integration, especially if you are dealing with Edge – Fog – Cloud continuum.
  • Reserve time for proper benchmarking and validation.
  • Consider earlier versions of software solutions. They may not have the latest features in place, but they are more stable and more predictable.
  • Allocate a task for data annotation since it is time-consuming and needs a lot of effort.
  • Active perception using re-enforcement learning: Aggressive word shaping may be required and also consider supervised approaches.
  • Interacting with clinical partners was challenging due to the Covid-19 pandemic.
  • Need to be up-to-date with regards to new technologies’ advancements and be able to quickly adapt/integrate them into the pipelines of the project.
  • Projects my face problems with respect to the time plan that was designed during the proposal shaping. If you do not have an HPC infrastructure ready in place when the project starts then keep in mind that it will need a significant amount of time to be ready for running/training your solutions. Other options should be considered in order not to delay the development of the project.