Frank Nielsen
Frank Nielsen works on computational and information geometry and their use in machine learning. He earned a doctorate in computational geometry in 1996. His research includes Bregman Voronoi diagrams, clustering, statistical divergences, and distances between probability models. With Alexander Soen, he described the pyBregMan library in 2024. The library implements operations on Bregman and Fisher-Rao manifolds and provides algorithms for statistics and machine learning. He has a research role at Sony Computer Science Laboratories in Tokyo.
Timeline
-
1996
Computational geometry training
Nielsen earned a master’s degree in parallel and distributed computing at ENS Lyon and a doctorate in computational geometry in 1996, prepared at Inria Sophia-Antipolis. His later research connects geometric methods with information science, visual computing, and machine learning.
-
2014
Clustering and statistical models
His 2014 work examines optimal interval clustering for Bregman clustering and statistical mixture learning. His work also addresses bounds on classification error.
-
2024
An information geometry library
Nielsen and Alexander Soen described pyBregMan in August 2024. The paper presents a Python library for generic operations on Bregman manifolds. It connects those spaces with pairs of dual potential functions and divergences, providing computational tools for a geometric approach to data.
-
2024
Probability models as geometric spaces
The pyBregMan paper also describes Fisher-Rao manifolds for categorical and multivariate normal distributions. Its example algorithms address statistics, machine learning, and information fusion.
Sources and credits
Photo: ELLIS profile; creator not stated. License: Not stated by source. Source: ellis.eu. Cropped to a square and resized. Ask us to remove or replace this photo.