A team of astronomers and computer scientists at the University of Hertfordshire, UK have taught a machine to ‘see’ astronomical images.
The technique, which uses a form of artificial intelligence called unsupervised machine learning, allows galaxies to be automatically classified at high speed, something previously done by thousands of human volunteers in projects like Galaxy Zoo.
The results are being presented at the UK National Astronomy Meeting in Llandudno, Wales, and the details of the algorithm are described in this paper.
The following are described in the paper published.
“They have demonstrated how a growing neural gas (GNG) can be used to encode the feature space of imaging data. When coupled with a technique called hierarchical clustering, imaging data can be automatically segmented and labelled by organising nodes in the GNG. The key distinction of unsupervised learning is that these labels need not be known prior to training, rather they are determined by the algorithm itself.
Importantly, after training a network can be be presented with images it has never ‘seen’ before and provide consistent categorisation of features. As a proof-of-concept , the team have demonstrated their algorithm using data from the Hubble Space Telescope ‘Frontier Fields‘: exquisite images of distant clusters of galaxies that contain several different types of galaxy.”
The Hubble Space Telescope is a space telescope that was launched into low Earth orbit in 1990 and remains in operation.
Hubble is the only telescope designed to be serviced in space by astronauts. After launch by Space Shuttle Discovery in 1990, four subsequent Space Shuttle missions repaired, upgraded, and replaced systems on the telescope.
The Hubble Space Telescope, a joint ESA and NASA project, has made some of the most dramatic discoveries in the history of astronomy. From its vantage point 600 km above the Earth, Hubble can detect light with ‘eyes’ 5 times sharper than the best ground-based telescopes and looks deep into space where some of the most profound mysteries are still buried in the mists of time.
In celebration of the Hubble Space Telescope’s 25th anniversary, NASA has released 25 of Hubble’s breathtaking and significant images.
Mr Hocking, who led the new work, said “The important thing about our algorithm is that we have not told the machine what to look for in the images, but instead taught it how to ‘see’.”
“A human looking at these images can intuitively pick out and instinctively classify different types of object without being given any additional information. They have taught a machine to do the same thing.”
“That could include ultrasound, micrographs, CAT scans, MRI—really any imaging data set where one might be looking for patterns. Again, the key thing is that this algorithm could search for very subtle features buried in the data that a human might miss,” Jim Geach, one of the researchers, told Gizmodo.
“Our aim is to deploy this tool on the next generation of giant imaging surveys where no human, or even group of humans, could closely inspect every piece of data. But this algorithm has a huge number of applications far beyond astronomy, and investigating these applications will be our next step,” concludes Geach.
This technique could also find application in other fields like medicine, where it could for example help doctors to spot tumours, and in security, to find suspicious items in airport scans.
