Sparse online self-organizing maps for large relational data

Madalina Olteanu (1) , Nathalie Vialaneix (2)

1 SAMM - Statistique, Analyse et Modélisation Multidisciplinaire (SAmos-Marin Mersenne)
2 MIAT INRA - Unité de Mathématiques et Informatique Appliquées de Toulouse

During the last decades, self-organizing maps were proven to be useful tools for exploring data. While the original algorithm was designed for numerical vectors, the data became more and more complex, being frequently too rich to be described by a fixed set of numerical attributes. Several extensions of the original SOM were proposed in the literature for handling kernel or dissimilarity data. Most of them use the entire kernel/dissimilarity matrix, which requires at least quadratic complexity and becomes rapidly unfeasible for 100 000 inputs, for instance. In the present manuscript, we propose a sparse version of the online relational SOM, which sequentially increases the composition of the prototypes.

Keywords: relational data; online relational SOM; sparse approximations

Domains: Statistics [stat]; Machine Learning [stat.ML]

Madalina Olteanu, Nathalie Vialaneix. Sparse online self-organizing maps for large relational data. WSOM 2016, Jan 2016, Houston, TX, United States. pp.27-37, ⟨10.1007/978-3-319-28518-4_6⟩. ⟨hal-01270710⟩