UK researchers are teaching computers to see and label galaxies using unsupervised machine learning. The group from the University of Hertfordshire, Hatfield, presents a novel unsupervised learning approach to automatically segment and label images in astronomical surveys.
Automation of this procedure will be essential as next-generation surveys enter the petabyte scale: data volumes will exceed the capability of even large crowd-sourced analyses. We demonstrate 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 organizing 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 categorization of features.
Teaching a Computer to ‘See’ Galaxies in Hubble Pics
Jul 15, 2015
Discuss (0)

The research team used a Tesla K40 and is working now to port parts of their code to CUDA in hopes to dramatically improve their performance.
Related resources
- GTC session: Generate Photorealistic Synthetic Data to Enhance AI Models with NVIDIA Omniverse Replicator
- GTC session: The Big Data Challenge of Radio Astronomy
- GTC session: Applying AI Weather Models With NVIDIA Earth-2
- NGC Containers: Microvolution
- NGC Containers: MATLAB
- SDK: IndeX - Amazon Web Services