Fencing AI for Enhanced Reliability in Geo Services

The increasing volume of Earth Observation (EO) and other scientific data presents challenges in extracting meaningful insights. While recent advances in AI/ML offer promising solutions, their effective application is hindered by complexity and the risk of applying models to inappropriate data, leading to inaccurate or misleading results.
FAIRgeo aims at (i) simplifying AI use on spatio-temporal Big Earth Data and (ii) investigates ways to restrict model application to its "comfort zone" of high accuracy (independent from the - usually unknown - training data), through techniques subsumed under "model fencing". This fencing will be applied automatically by the server, on models uploaded by users.
The open OGC/ISO Coverages and WCPS standards form the conceptual basis, implementation is done using the pioneer geo datacube engine, rasdaman.