Apologies for cross posting
--------------------------------- SEMINARIO DI MATEMATICA APPLICATA
Giovedi 25 Febbraio 2016, ore 10.30 Aula Rappresentanza, Dipartimento di Matematica (Via Saldini 50, Milano)
Robust statistical techniques for the analysis of high dimensional gene expression data
Dr. Aurora Torrente Gregorio Millan Institute Escuela Politecnica Superior Universidad Carlos III de Madrid
ABSTRACT Microarray and Next Generation Sequencing experiments provide data on the expression level of thousands of genes simultaneously, suggesting several serious challenges in the pre-processing, management and analysis of these high dimensional datasets. In this talk, we will describe some robust statistical tools, based on the concept of functional data depth, which measures the centrality of a function within a sample. These techniques include, for example, the scale curve, as a way to visualise the dispersion of a set of multivariate observations, and supervised classifications techniques, to assign new observations to one of a given set of (known) groups. When these tools make use of the so-called Modified Band Depth, adapted to high-dimensional data, they prove to be very efficient and of low computational cost, specially when a suitable implementation is considered. The use of these techniques will be illustrated on real gene expression data.
La S.V. e` cordialmente invitata.
-----------------------------------------------------------------------
Home page del seminario: http://www.mat.unimi.it/users/veeser/sma/ Il seminario e` organizzato da: - Dipartimento di Matematica dell'Universita` di Milano, - Scuola di Dottorato in Scienze Matematiche.