PROCEEDINGS IPMU '08
Identification of a fuzzy measure with an L1 entropy
Christophe Labreuche.
The aim of the paper is to propose
a linearized entropy of a fuzzy measure.
This entropy is used as an optimization
functional for the identification
of a fuzzy measure from learning
data. The problem of identifying
a fuzzy measure can thus be transformed
into a linear program while
keeping the good properties of the
entropy.
PDF full paper |