Towards a partitioning of the input space of Boolean networks: Variable selection using bagging

Abstract

In this paper we present an algorithm that allows to select the input variables of Boolean networks from incomplete data. More precisely, sets of input variables, instead of single variables, are evaluated using mutual information to find the combination that maximizes the mutual information of input and output variables. To account for the incompleteness of the data bootstrap aggregation is used to find a stable solution that is numerically demonstrated to be superior in many cases to the solution found by using the complete data set all at once. © 2009 ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering.