Machine Learning for Matching Radio and Optical Telescope Data

Exhibitor: Dr Marcus Gallagher

Research Group: Complex and Intelligent Systems

Industry Sector: Scientific / Research Services

The Problem : Radio and optical telescopes have both done complete surveys of the southern sky. The radio telescopes’ galaxy detections have a large positional error so that there are multiple optical candidates for each radio source. Which optical counterpart (left) matches which radio counter part (right)? Traditionally this has been done by eye – but can we automate the process? In more general terms this is a problem of matching two disparate datasets.

The Solution: Use a machine learning technique to learn the associations between optical and radio sources.