A new generation of astronomical surveys will produce petabytes of data requiring new automated methods to categorise astronomical objects. We propose an unsupervised learning approach to automatically categorise galaxies within multi-wavelength Hubble Space Telescope images. Using growing neural gas and agglomerative clustering we demonstrate the efficient categorisation of galaxy structure into star-forming and passive regions and present the key steps required to produce catalogues of galaxies. We suggest that the algorithm has potential as a tool in the automatic analysis and data mining of the next generation imaging and spectral surveys, and could also find application beyond astronomy.

Date: 04/09/2015
Time: 16:00
Location: LB252