Neural networks and fuzzy systems: a dynamical systems approach to machine intelligence
Neural networks and fuzzy systems: a dynamical systems approach to machine intelligence
Self-organization as an iterative kernel smoothing process
Neural Computation
Journal of the American Society for Information Science
A corpus-based approach to comparative evaluation of statistical term association measures
Journal of the American Society for Information Science and Technology
Self-Organizing Maps
Modern Information Retrieval
Multivariate Descriptive Statistical Analysis
Multivariate Descriptive Statistical Analysis
Redefining Clustering for High-Dimensional Applications
IEEE Transactions on Knowledge and Data Engineering
A Comparative Study on Feature Selection in Text Categorization
ICML '97 Proceedings of the Fourteenth International Conference on Machine Learning
Self Organizing Map and Sammon Mapping for Asymmetric Proximities
ICANN '01 Proceedings of the International Conference on Artificial Neural Networks
Visualizing asymmetric proximities with SOM and MDS models
Neurocomputing
Self organization of a massive document collection
IEEE Transactions on Neural Networks
Self-organizing maps, vector quantization, and mixture modeling
IEEE Transactions on Neural Networks
Functional analysis techniques to improve similarity matrices in discrimination problems
Journal of Multivariate Analysis
Hi-index | 0.00 |
Self Organizing Maps (SOM) are useful tools to discover the underlying structure of high dimensional data. However the algorithms proposed in the literature rely on the use of symmetric measures such as the Euclidean. Therefore when asymmetry arises they fail to reflect accurately the object proximities and the resulting maps become often meaningless. This is a serious drawback for several applications such as text mining in which the object relations are strongly asymmetric. In this paper, we propose two variants of the original SOM algorithm that are able to deal successfully with asymmetric relations. The algorithms are tested using real document collections, and the performance is reported using appropriate measures. The asymmetric algorithms improve significantly the maps generated by their symmetric counterpart.