Abstract
In recent decades; several machine learning methods based on Pot-trial Concept, Analysis have been proposed. The learning process is based on the construction of the mathematical structure of the Galois lattice. Two major limits characterize these methods. First, most of them are limited to the binary data. processing. Second; the exponential complexity of a Galois lattice generation limits their fields of application. In this paper; we consider the Boosting of classifiers; which is an adaptive approach of classification. We propose the Boosting of classifiers based on Nominal Concepts. This method builds part of the lattice including the best concepts (pertinent; concepts). It is distinguished front the other methods based on Formal Concept Analysis by its ability to handle nominal data. The discovered concepts are called Nominal Concepts and they are used as classification rules. The comparative studies and the experimental results carried out; prove the interest; of this method compared to those existing in literature.