The DLT priority sampling is essentially optimal.

Published in: STOC (2006)

Keyphrases

optimal solution
random sampling
neural network
worst case
dynamic programming
computational complexity
lower bound
active learning
response time
sample size
monte carlo
sampling strategy