UNIVERSITY OF HERTFORDSHIRE
		 COMPUTER SCIENCE RESEARCH COLLOQUIUM

			      presents

	     "Causal Information Flows and the Common 
                          Cause Principle"

		          Prof. Nihat Ay
    (Max Planck Institute for Mathematics in the Sciences, Germany  
                       &  Santa Fe Institute, USA) 


		     19 June 2008 (Thursday)

		      Seminar Room F315 
                  Hatfield, College Lane Campus
			   3 - 4 pm

	       Coffee/tea and biscuits will be available.

		    Everyone is Welcome to Attend


Abstract:


    Commonly used notions of information flow are based on
mutual information which, however, quantifies general stochastic
dependence without reference to the causal nature of information
flow. In order to specify associations due to causal effects,
interventional conditioning has to be applied. This idea leads to a
new information flow measure which Daniel Polani and I recently
introduced.  Based on this measure, I discuss refinements of the
so-called common cause principle. A first version of it relates
general stochastic dependence among a set of random variables to a sum
of information flows through "essential channels". A further analysis
along this line, mainly developed by Bastian Steudel, yields
structural constraints imposed by high stochastic dependence. More
precisely, if the stochastic dependence exceeds a threshold value
common causes of higher order can be inferred.

--------------------------------------------------
Hertfordshire Computer Science Research Colloquium
http://homepages.feis.herts.ac.uk/~nehaniv/colloq