Glad to share a freshly submitted paper (now under peer review) from the VASCO team, led by Prof.

Glad to share a freshly submitted paper (now under peer review) from the VASCO team, led by Prof. Steve Bruehl. We use machine learning to separate likely real transients from plate defects in a sample of 107,875 events. If correlations were spurious, they should weaken as artifacts are removed. Well...they don’t. Instead, the correlations strengthen as the sample gets cleaner. We see: (1) Stronger Earth shadow deficit (2) Stronger nuclear-test correlation (3) Highest in the highest-probabili...