Colloquium Computing Science

 

 

Date:                            Monday, February 25 2008      

Speaker:                      Fabrice Rossi

Room and time:           5161.0267 (Bernoulliborg), 16.00

 

 

 

Title:    An Introduction to Functional Data Analysis

 

Abstract:


Curve or functional data are natural in many application fields, for instance
those that involve time series analysis or some kind of spectra. In these
applications, each observation corresponds to a mapping between some
conditions (that might be implicit) and the observed response. In practice,
the functional observation are sampled. This generally leads to a vector
representation to which standard multivariate tools could be applied The
direct use of classical models for this type of data faces however several
difficulties: when the underlying functions are smooth, the sampling generates
high dimensional vectors with highly correlated coordinates. As a consequence,
classical methods lead to ill-posed problems, both on a theoretical point of
view (when working in functional spaces that have infinite dimension) and on a
practical one (when working with the sampled functions). The goal of
Functional Data Analysis (FDA) is to use, in data analysis algorithms, the
underlying functional nature of the data in order to circumvent both
problems.

This talk will introduce the main concepts of functional data analysis and
summarize important results, both on a practical point of view (how to model
functional data for real world applications) and on a theoretical one (how to
guarantee the consistency one the methods while dealing with infinite
dimensional spaces).

 

 

Colloquiumcoördinator is Prof.dr. M. Aiello, Dr. M. Biehl (e-mail: M.Biehl@cs.rug.nl)

www.cs.rug.nl/~biehl/Coll