Automated Classification of 2000 Bright IRAS Sources

Abstract

An Artificial Neural Network (ANN) has been employed using a supervised back-propagation scheme to classify 2000 bright sources from the Calgary database of IRAS (Infrared Astronomy Satellite) spectra in the wavelength region of 8-23 microns. The data base has been classified into 17 pre-determined classes based on spectral morphology. We have been able to classify more than 80 percent of the 2000 sources correctly at the first instance. The speed and robustness of the scheme will allow us to classify the whole of LRS database, containing more than 50,000 sources in the future.

26 pages, To appear in ApJS after July 2004

Cited by in corpus (10)

• Data Mining and Machine Learning in Astronomy
• IRAS-based whole-sky upper limit on Dyson Spheres
• Characterisation of Galactic carbon stars and related stars from Gaia-EDR3
• Comparing Pattern Recognition Feature Sets for Sorting Triples in the FIRST Database
• The VLTI/MIDI view on the inner mass loss of evolved stars from the Herschel MESS sample
• Automated Classification of Sloan Digital Sky Survey (SDSS) Stellar Spectra using Artificial Neural Networks
• Active galactic nuclei synapses: X-ray versus optical classifications using artificial neural networks
• Infrared Emission from the Composite Grains: Effects of Inclusions and Porosities on the 10 and 18 $\mu m$ Features
• Composite Circumstellar Dust Grains
• A 3D Automated Classification Scheme for the TAUVEX data pipeline