Monkeys at Typewriters: Using Unsupervised Learning for Automatic Classification of Galaxies and Industrial Innovation

School of Physical Sciences Seminar

Tuesday 11^th June at 11.00 a.m.

Robert Hooke Seminar Room

Coffee available from 10.30 a.m.

Dr Ashley Spindler, (University of Hertfordshire)

As we enter the age of petabyte data, new and innovative data analysis techniques are required. Machine Learning comes in many different flavours, and provides a backbone for the analysis of large datasets from next generation observatories. Most astronomical research into machine learning focuses on supervised techniques, such as neural networks, however, unsupervised learning could be useful for situations where training data is not available. In this talk I will discuss the techniques being used to study the automatic classification of galaxies in the Hubble Frontier Fields using unsupervised learning techniques, such as self-organising maps, t-distributed stochastic neighbour embedding and hierarchical clustering. I will also discuss how these techniques also open the door to industrial partnership and innovation in remote sensing applications, such as agricultural imaging and SONAR.

Please remember to sign in on the attendance sheet.