Decomposing probability distributions on structured individualsPeter A. Flach, Nicolas Lachiche, Decomposing probability distributions on structured individuals. Proceedings of the ECML2000 workshop on Dealing with Structured Data in Machine Learning and Statistics. Paula Brito, Joaquim Costa, Donato Malerba, (eds.), pp. 33–43. May 2000. PDF, 68 Kbytes.
Naive Bayesian classifiers are very successful in attribute-value representations. However, it is not clear how the decomposition of the probability distributions on attribute-value tuples underlying those classifiers can be applied on structured individuals, for instance sets of tuples as in the multiple instance problem. This paper presents a decomposition of probability distributions on structured individuals. It shows how it results in a propositionalisation of the data guided by the decomposition of the structure of the individual. This is illustrated by using the first-order naive Bayesian classifier 1BC to perform the decomposition of probability distributions on structured individuals.