Batch self-organizing maps based on city-block distances for interval variables

Francisco de A. T. de Carvalho (1) , Patrice Bertrand (2) , Filipe M. de Melo (1)

1 CIn - Centro de Informatica UFPE [Recife]
2 CEREMADE - CEntre de REcherches en MAthématiques de la DEcision

The Kohonen Self Organizing Map (SOM) is an unsupervised neural network method with a competitive learning strategy which has both clustering and visualization properties. Interval-valued data arise in practical situations such as recording monthly interval temperatures at meteorological stations, daily interval stock prices, etc. Batch SOM algorithms based on adaptive and non-adaptive city-block distances, suitable for objects described by interval-valued variables, that, for a fixed epoch, optimizes a cost function, are presented. The performance, robustness and usefulness of these SOM algorithms are illustrated with real interval-valued data sets.

Mots clés: Self-organizing maps; Interval-valued data; City-block distances; Adaptive distances

Domaines: Statistiques [math.ST]; Théorie [stat.TH]; Apprentissage [cs.LG]

Francisco de A. T. de Carvalho, Patrice Bertrand, Filipe M. de Melo. Batch self-organizing maps based on city-block distances for interval variables. 2012. ⟨hal-00706519⟩