Talk: Causally Inspired Approaches to Variable Selection, Molecular Signature Identification, and Integrative Analysis of Heterogeneous Datasets

Prof. Ioannis Tsamardinos (Head of the Bioinformatics Laboratory at ICS‐FORTH in Greece) will serve as Antti Hyttinen's PhD defence opponent, and during his visit he will give a talk

 

Tuesday, May 7th, at 14.15 in C222 (Exactum, Kumpula)

 

on topics related to the learning of Bayesian networks and feature selection, with applications in biology.

 

TITLE: Causally Inspired Approaches to Variable Selection, Molecular Signature Identification, and Integrative Analysis of Heterogeneous Datasets

 

ABSTRACT: Inspired by our notion of causality, several graphical probabilistic formalisms and corresponding discovery algorithms have been proposed such as Bayesian Networks and Maximal Ancestral Graphs. The success of these models in probabilistic reasoning is partly the reason for the recent Turing Award to one of their founders, Prof. Judea Pearl. Several successful applications of computational causal discoveries that will be presented already exist in biology. In this talk, examine the connection of these causally-inspired models to solving long-standing important data-analysis problems. Namely, we examine the causal perspective to (a) the problem of variable selection, a.k.a. feature selection and molecular signature identification in the context of biology and (b) the problem of the co-analyses of several heterogeneous datasets in the context of prior causal knowledge. Datasets can be heterogeneous in the sense that they measure different (but overlapping) variable sets or under different experimental conditions.

 

About the Speaker:

 

Ioannis Tsamardinos is the Head of the Bioinformatics Laboratory at ICS‐FORTH, an Assistant Professor at the Department of Computer Science at University of Crete, and an Adjunct Assistant Professor at the Department of Biomedical Informatics at Vanderbilt University. He received his Ph.D. in 2001 from the Intelligent Systems Program of the University of Pittsburgh and worked as an Assistant Professor at the Dept. of Biomedical Informatics at Vanderbilt University between 2001 and 2006. Ioannis has over 60 publications in international journals, conferences, and books in the fields of Machine Learning, Artificial Intelligence and Bioinformatics. Research interests include developing methods for feature selection and causal discovery and applications to biological data, introducing some of the first methods for local causal learning and integrative causal analysis. Distinctions with colleagues and students include the best performance in one of the four tasks in the recent First Causality Challenge Competition, a Gold Medal in the Student Paper Competition in MEDINFO 2004, the Outstanding Student Paper Award in AIPS 2000, the NASA Group Achievement Award for participation in the Remote Agent team and others.