Fuzzy techniques for coffee flavour classification
Résumé
This paper deals with the differentiation of pure and mixed samples of commercialized coffees. Fuzzy c-means and k-nearest neighbours algorithms are used to discriminate pure arabica and robusta samples by treating the initial feature vectors. When samples sets including mixtures are analysed, these techniques fail to differentiate efficiently this category of samples. A data processing by means of a discriminant analysis provides two new features which are used for classification purposes using the k-nearest neighbours algorithm
Origine : Fichiers produits par l'(les) auteur(s)
Loading...