Seminar (SS 2017)

Statistics of non-euclidean data

Preliminary discussion:

Tuesday, April 25th, 2017, 15:00,
MΛTHEMΛTIKON, INF 205, 4th floor, room 4.414

Time and location of the seminar:

The seminar will be held as a Blockseminar on:

June 2nd, 2017, 09:15-17:00, MΛTHEMΛTIKON, INF 205, 4th floor, R 4.414

Contact:

Christof Schötz <schoetz[at]math.uni-heidelberg.de>

Enno Mammen <mammen[at]math.uni-heidelberg.de>

Jan JOHANNES <johannes[at]math.uni-heidelberg.de>

Questions, please directly by email to Christof Schötz or by using the contact form.

Language:

The seminar will be hold in English.

Area:

Applied Mathematics, Stochastics

Please register for the seminar by email to Christof Schötz.

Description of the seminar:

In this seminar we will discuss approaches to probability theory and statistics for data that are not Euclidean, i.e., for data that cannot be represented as elements of Rn. If the data live in a metric space, a natural question is how one defines a mean. The Fréchet mean (also called barycenter) gives one possible answer: it defines the mean of elements of a metric space as the value that minimizes the sum of squared distances to these elements. Based on this notion of mean, one can try to reproduce results and statistical methods from the Euclidean case in more abstract settings. In this seminar we want to present the basic concepts needed for probability theory in metric spaces, explore some resulting propositions and implications thereof, and see how these can be used for statistics.

Requirements:

The seminar is directed to Master students who want to specialize in statistics.

References:

See Wiki .