Peering Into the Abyss: Machine Learning Enhances M87 Black Hole Image
AI provides better black hole images.
The famous picture of the black hole at the center of M87 has received its first official remodeling with the help of AI. The new drawing showcases a central area that is larger and darker, enveloped by gas shaped like a “donut.” The data came from the Event Horizon Telescope (EHT) collaboration a few years ago.
PRIMO, the new technique, helps with achieving the maximum resolution of the current array.
PRIMO was developed by Lia Medeiros, Dimitrios Psaltis and others.
It is a new way to the challenging mission of creating images from EHT observations. It compensates for the lacking information about the object being monitored, which is needed to produce the image that would have been seen through a single gigantic radio telescope.
PRIMO is based on dictionary learning, which enables computers to generate rules based on big sets of training material (if a computer is given a series of different banana pictures—with enough training—it can determine if an unknown image is or not a banana).
PRIMO has helped computers with the analysis of over 30,000 high-fidelity simulated images of black holes accreting gas. The duplicates covered a different kinds of models for how the black hole gains matter, looking for routine patterns in the structure of the forms.
The new AI-based methods give a worthwhile chance for understanding the physics of black hole.
AI Catalog's chief editor