@inproceedings{2815b2855bd94c7eaba59fd8a376a9f9,
title = "PPoSOM: A multidimensional data visualization using probabilistic assignment based on polar SOM",
abstract = "A new algorithm named probabilistic polar self-organizing map (PPoSOM) is proposed. PPoSOM is a new variant of Polar SOM which is constructed on 2-D polar coordinates. Data weight and feature are represented by two variables that are radius and angle. The neurons on the map are set as data characteristic benchmarks. Projected data points are trained to get close to the neurons with the highest similarities, while weights of neurons are updated by a probabilistic data assignment method. Thus, not only similar data are gathered together, data characteristics are also reflected by their positions on the map. Our obtained results are compared with conventional SOM and ViSOM. The comparative results show that PPoSOM is a new effective method for multidimensional data visualization. {\textcopyright} 2009 Springer-Verlag Berlin Heidelberg.",
keywords = "Probabilistic polar SOM (PPoSOM), SOM, ViSOM, Visualization",
author = "Yang Xu and Lu Xu and Chow, {Tommy W. S.} and Fong, {Anthony S. S.}",
year = "2009",
doi = "10.1007/978-3-642-10677-4_24",
language = "English",
isbn = "3642106765",
volume = "5863 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "212--220",
booktitle = "Neural Information Processing",
address = "Germany",
note = "16th International Conference on Neural Information Processing (ICONIP 2009) ; Conference date: 01-12-2009 Through 05-12-2009",
}