A crucial issue in dissimilarity-based classification is the choice of the representation set. In the small sample case, classifiers car pable of a good generalization and the injection or addition of extra information allow to overcome the representational limitations. In this paper, we present a new approach for enriching dissimilarity representations. It is based on the concept of feature lines and consists in deriving a generalized version of the original dissimilarity representation by using feature lines as prototypes. We use a linear normal density-based classifier and the nearest neighbor rule, as well as two different methods for selecting prototypes: random choice and a length-based selection of the feature lines. An important observation is that just a few long feature lines are needed to obtain a significant improvement in performance over the other representation sets and classifiers. In general, the experiments show that this alternative representation is especially profitable for some correlated datasets. © Springer-Verlag Berlin Heidelberg 2007.
CITATION STYLE
Orozco-Alzate, M., Duin, R. P. W., & Castellanos-Domínguez, C. G. (2007). Generalizing dissimilarity representations using feature lines. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 4756 LNCS, pp. 370–379). https://doi.org/10.1007/978-3-540-76725-1_39
Mendeley helps you to discover research relevant for your work.