Planets that do not transit are very difficult for Kepler to find. The authors of today’s paper are not intimidated by that and find 60 non-transiting Hot Jupiters unknowingly detected by Kepler anyway.
Neural networks are very good at their jobs. We may be starting to understand how.
Machine learning has been covered in lots of previous Astrobites. But today’s paper is about machine learning off training wheels: a fully automated “unsupervised” method of classifying galaxy morphology.
New astronomy is published every day, but are we actually learning from it?
Sometimes, computers are too slow to get the job done. These researchers found a faster way to predict whether planetary systems are stable.
How an artificial neural network can be trained to classify stars into spectral types, using only a single broad-band image.