Browsing by Subject "Sequential hypothesis testing"

Now showing 1 - 1 of 1

Finding good enough coins under symmetric and asymmetric information
(2017-12-08) Pattabiraman, Srilakshmi; Caramanis, Constantine
We study the problem of returning m coins with biases above 0:5. These good enough coins that are returned by the agent should be acceptable to the authority by meeting the authority's Family Wise Error Rate constraint. We design adaptive algorithms that invoke Sequential Probability Ratio Test to find these good enough coins. We consider scenarios that differ in terms of the information available about the underlying Bayesian setting. The symmetry or asymmetry of the underlying setup, i.e., the difference between what the agent and the authority know about the underlying prior and the support, presents different challenges. We also make notes on the algorithms' sample complexity.