[LCT01] Probabilistic Topological Map and Binary data
Conférence Internationale avec comité de lecture :
ICANN2001,
January 2001,
motcle:
Résumé:
The Self Organizing Map (SOM) proposed by Kohonen is a well known neural
model which provides both quantization and clustering of the observation
space. In this paper, we adapt the Bernoulli mixture approach to the model of binary topological map and show that using a
probabilistic formalism gives rise to better quantization process and
classification performances.
Commentaires:
Soumis à ICANN2001
Collaboration:
UPMC