SWAPE jointly learns parameters and sparse sample weights
Alternating majorize-minimize updates cover convex and difference-of-convex losses while down-weighting anomalous points.
Stories per day · 30 days
1 stories, newest first · one card per story