Event Details

Physics-Informed Statistical Learning for Spatial and Functional Data

Physics-Informed Statistical Learning for Spatial and Functional Data

This course introduces a family of physics-informed statistical learning methods for spatial, spatio-temporal, and functional data. These methods incorporate roughness penalties based on differential operators, encoding information about the underlying physical processes and the geometry of the observation domain. The framework accommodates data defined over multidimensional domains with complex geometries, including non-convex planar regions, curved surfaces, non-convex volumes, and linear networks. Numerical discretization techniques based on unstructured meshes provide considerable modeling flexibility, enabling the representation of highly localized signals, strong anisotropies, and non-stationary patterns.

The course will combine methodological foundations with applications in environmental and life sciences, illustrating the potential of physics-informed statistical learning for the analysis of complex data. Hands-on sessions will introduce the open-source fdaPDE 2.0 library (https://github.com/fdaPDE), with interfaces in R and Python.

---



Schedule



Tuesday 24^th November

Masterclass session 1

10:00-12:00 Management School LT4

Tuesday 24^th November

Masterclass session 2

14:00-16:00 PSC LT (A54)

Wednesday 25^th November

Masterclass session 3

10:00-12:00 PSC LT (A54)