UH Biocomputation Group - Astronomy

The Science of Variable Stars and Transient Sources

2017-11-30T11:41:30+00:00

Philip will be presenting a talk on 'The Science of Variable Stars and Transient Sources'.

Many astronomical sources vary in brightness, for a wide variety of reasons, e.g. pulsating stars, exploding stars and quasars. The 1-dimensional time series contains a wealth of physical information that has been used to measure the size of the universe, to detect and characterise new solar systems and to study the structure and history of our Milky Way galaxy. Major new "big data" time domain projects such as VVV, LSST and TESS are seeking to further advance the field, requiring us to develop new methods to classify the vast number of variable sources and search for new types of behaviour. Astronomical time series datasets tend to be unevenly sampled and to have non-uniform uncertainties, influenced by correlated noise. Philip will introduce the field, outline some of the existing tools developed by astronomers to analyse periodic and non-periodic variables and indicate some of the new approaches being developed.

Date: 01/12/2017
Time: 16:00
Location: LB252

Mining Hubble Space Telescope images

2017-03-24T08:03:22+00:00

We present a unsupervised machine learning technique to explore large sky surveys produced by imaging telescopes. Distinct from previous approaches this technique requires no pre-selection of target galaxy type; instead it automatically identifies objects that are similar. We apply the technique to the five Hubble Space Telescope CANDELS fields and compare the machine-based classifications to human-classifications from the Galaxy Zoo: CANDELS project. We find that although there is not a direct mapping between Galaxy Zoo and our hierarchical labelling, there is a good level of concordance between human and machine classifications. A catalogue and galaxy similarity search is provided for the CANDELS fields at http://www.galaxyml.uk/

Date: 24/03/2017
Time: 16:00
Location: LB252

Automated morphological classification of galaxies using machine learning techniques

2016-07-14T11:42:14+01:00

The predominant method to analyse the morphology of galaxies is to use human visual inspection. The Galaxy Zoo project has successfully scaled this process to incorporate crowd sourced inspection and has released detailed classifications of hundreds of thousands of galaxies. However, the next generation of telescopes are now being designed and built. An optical telescope called the Large Synoptic Sky Telescope, for example, is due to come online in 2019 and will image half the sky every few days to catalogue ~40 billion objects. This is beyond the capability of even the significant crowd sourced resources of Galaxy Zoo. Therefore, astronomers are now investigating automated solutions using machine learning. In this talk I outline the key issues that need to be resolved in order to automate galaxy morphological analysis in large surveys. I describe existing automated systems that analyse galaxies, and I describe the results of my work using unsupervised machine learning approaches to characterize galaxies.

Date: 15/07/2016
Time: 16:00
Location: LB252