Navarro Silvana G.

Abstract Reference: 30202
Identifier: P1.21
Presentation: Poster presentation
Key Theme: 1 Reduction and Analysis Algorithms for Large Databases and Vice-versa

Automatic spectral classification of galaxies using SPITZER data

Authors:
Navarro Silvana G., Guzmán Violeta, Corral Luis J., Dafonte Carlos, Rodríguez Alejandra

In this work we present the analysis of the spectral sample obtained from the Spitzer Infrared Spectrograph (IRS) . We applied unsupervised neural networks: competitive (CNN) and self organized maps (SOM), to the sample of 747 galaxy spectra. All of them were obtained from the central part of the galaxies. The redshift (z) of the sample galaxies is between 0.0001 and 0.3. All are high resolution spectra (R ~ 600) obtained with the Short-High (9.9-19.6 μm) and Long-High (18.7-37.2μm) modules of IRS. We obtained an automatic classification on 17 groups with the CNN, and we compare the results with those obtained with SOM. The obtained classification with both methods are consistent. We are analyzing the physical properties of the galaxies in each group to determine what type of objects dominates each one and to confirm if the classification could be extended to a larger database.