Samarbeide med internasjonale partnere, delta i prosjektmøter og å skrive selve EU søknaden Selve prosjektet: The EXSCALE project will stand on three pillars: 1. First, a resource- and location-aware computing toolkit that implements dynamic resource management to facilitate deployment on resources of arbitrary architectures (CPU, GPU, FPGA, ASIC) at any network levels (fog, edge, cloud, dedicated). 2. Second, novel machine learning algorithms for distributed unsupervised online learning, upgrades to existing algorithms towards excellent multi-node scaling, and exploitation of recent advances in, for example, Deep Learning, to extract new insights from extreme-scale distributed streams. 3. Third, tools for automated critical-time decision support and visualization even in the face of extreme numbers of high-velocity data sources. While EXSCALE will be completely generic, its development will be driven by use-cases from three industrial sectors: energy, finance, and remote/moving sensing. These impose realistic, demanding, and critical requirements on the core aspects of deployment, processing, and decision making. They there therefore uniquely suited to validate the system. The EXSCALE project is structured into 9 work packages (WPs) with the following responsibilities: # Title Description 1 Project Management 2 Use Case Maturation; Mature use-cases into concrete pilots. Derive their needs and requirements wrt the overall platform. 3 Responsible AI; Ensure compliance with ethical AI guidelines. 4 Architecture; Core platform development. 5 Data Handling & Sharing; Development of scalable data fusion. 6 Extreme Scale AI; Development of scalable analytics and AI systems. 7 Automated Decision Making; Development of scalable automated decision making. 8 Benchmarking & Demo; Evaluation of the platform components (WP4-WP7) with regards to the use-case needs and requriemnts from WP2. 9 Exploitation; Coordination of exploitation activites.
Project leader: Kenneth Hauglund
Institution: SCIENCE AND TECHNOLOGY AS