Information Geometry: applications to signal processing and machine learning

Abstract

This project concerns study and research in the field of Information Geometry, which analyses the intrinsic geometry of families of probability distributions and statistical models, aiming particularly at applications in data processing and data analysis. Methods from this field, which are based on differential/Riemannian geometry, have been recently used in different contexts. The objective of this project is to study the foundations and contributions to this subject, and to research new efficient approaches and algorithms in applications to signal processing and machine learning.