inria-00326722, version 1

Learning Pullback Metrics for Linear Models

Fabio Cuzzolin ¹

The 1st International Workshop on Machine Learning for Vision-based Motion Analysis - MLVMA'08 (2008)

• 1:  Oxford Brookes University
• http://www.brookes.ac.uk/
  Oxford Brookes University Headington Campus, Gipsy Lane, Oxford OX3 0BP, UK United Kingdom

Bibliographic reference

• Type of document: Congres communications
• Domain: Computer Science/Computer Vision and Pattern Recognition
• Title: Learning Pullback Metrics for Linear Models
• Abstract: In this paper we present an unsupervised differential-geometric approach for learning Riemannian metrics for dynamical models. Given a training set of models the optimal metric is selected among a family of pullback metrics induced by the Fisher information tensor through a parameterized diffeomorphism. The problem of classifying motions, encoded as dynamical models of a certain class, can then be posed on the learnt manifold. Experimental results concerning action and identity recognition based on simple scalar features are shown, proving how learning a metric actually improves classification rates when compared with Fisher geodesic distance and other classical distance functions.
• Full text language: English
• Publication date: 2008
• Audience: international
• Conference title: The 1st International Workshop on Machine Learning for Vision-based Motion Analysis - MLVMA'08
• Conference city: Marseille
• Country: France
• Conference date: 2008-10

Attached file list to this document: 

• inria-00326722, version 1
• http://hal.inria.fr/inria-00326722
• oai:hal.inria.fr:inria-00326722
• From: Peter Sturm <>
• Submitted on: Sunday, 5 October 2008 12:44:48
• Updated on: Monday, 6 October 2008 09:40:12