HSC: A multi-resolution clustering strategy in Self-Organizing Maps applied to astronomical observations

摘要: This work presents a strategy for the classification of astronomical objects based on spectrophotometric data and use unsupervised neural networks statistical algorithms. Our constitutes an essential part preparation phase automatic parameterization algorithms that are to be collected by Gaia satellite European Space Agency (ESA), whose launch is foreseen spring 2012. The proposed algorithm hierarchical structure composed various tree-structured SOM networks. possible (stars, galaxies, quasars, multiple objects, etc.) basically consists in iterative segmentation inputs space ensuing generation initial classifications increase precision means refining process. Apart from providing classification, our technique also measures quality segments which it cannot determine whether or not they belong pre-established class (outliers).