In this paper a Self-Organizing Map (SOM) robust to the presence of outliers, the Smoothed SOM (S-SOM), is proposed. S-SOM improves the properties of input density mapping, vector quantization, and clustering of the standard SOM in the presence of outliers by upgrading the learning rule in order to smooth the representation of outlying input vectors onto the map. The upgrade of the learning rule is based on the complementary exponential distance between the input vector and its closest codebook. The convergence of the S-SOM to a stable state is proved. Three comparative simulation studies and a suggestive application to digital innovation data show the robustness and effectiveness of the proposed S-SOM. Supplementary materials for this article are available.
Smoothed self-organizing map for robust clustering / D'Urso, P.; De Giovanni, L.; Massari, R.. - In: INFORMATION SCIENCES. - ISSN 0020-0255. - 512:(2019), pp. 381-401. [10.1016/j.ins.2019.06.038]
Smoothed self-organizing map for robust clustering
D'Urso P.;De Giovanni L.;Massari R.
2019
Abstract
In this paper a Self-Organizing Map (SOM) robust to the presence of outliers, the Smoothed SOM (S-SOM), is proposed. S-SOM improves the properties of input density mapping, vector quantization, and clustering of the standard SOM in the presence of outliers by upgrading the learning rule in order to smooth the representation of outlying input vectors onto the map. The upgrade of the learning rule is based on the complementary exponential distance between the input vector and its closest codebook. The convergence of the S-SOM to a stable state is proved. Three comparative simulation studies and a suggestive application to digital innovation data show the robustness and effectiveness of the proposed S-SOM. Supplementary materials for this article are available.File | Dimensione | Formato | |
---|---|---|---|
Information Sciences (D'Urso et al., 2019).pdf
solo gestori archivio
Tipologia:
Documento in Post-print (versione successiva alla peer review e accettata per la pubblicazione)
Licenza:
Creative commons
Dimensione
979.58 kB
Formato
Adobe PDF
|
979.58 kB | Adobe PDF | Contatta l'autore |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.