In learning theory there is focus on "non-asymptotic" results. Instead of only showing that our method converges on the right answer in the limit of infinite da

In learning theory there is focus on "non-asymptotic" results. Instead of only showing that our method converges on the right answer in the limit of infinite data, we must show how fast it converges.