SPS Seminar - Using machine learning to identify gravitational lenses candidates in Hubble data

When:  Thursday 29^th September at 14.00

Where:  Microsoft Teams – Online/Robert Hooke Seminar Room

Speaker:  Joshe Wilde (OU)

Hosted by: Hugh Dickinson

Abstract:

In the near future, the Euclid space telescope will discover 10,000s of new gravitational lenses. We will need techniques such as machine learning to identify all of these new gravitational lenses. I have spent my PhD developing machine learning based approaches to identify gravitational lenses in simulated Euclid-like images. I will give a brief introduction of gravitational lensing and modern machine learning methods, followed by the approaches I have developed, highlighting how these methods detect gravitational lenses. I will then present the results of these methods on Hubble data.  Finally, concluding with my next steps and lessons learnt.

Bio:

Josh Wilde is a final year PhD student at The Open University. His research has focused on applying machine learning to the identification of gravitational lenses using simulated Euclid images. Josh has developed a Convolutional Neural Network (CNN) which has successfully identified gravitational lenses showing the features the CNN focuses on.