Yeah - SVM were considered to be optimal, then they got smashed by deep networks, now people keep bringing them back and saying "they are just as good" and yet

Yeah - SVM were considered to be optimal, then they got smashed by deep networks, now people keep bringing them back and saying "they are just as good" and yet deep networks keep being used to do all the breakthrough work.