ar(k) stuff, sure. that's old news. i would expect the newfangled stuff to be good at 0-shot learning of pre-event signatures spread across multiple series, at

ar(k) stuff, sure. that's old news. i would expect the newfangled stuff to be good at 0-shot learning of pre-event signatures spread across multiple series, at a minimum.