> A neural net can approximate any continuous function. Can't a RF or boosted regression tree theoretically do the same and then some (approximate discontinuou

> A neural net can approximate any continuous function. Can't a RF or boosted regression tree theoretically do the same and then some (approximate discontinuous functions)?