A New Bound on the Cumulant Generating Function of Dirichlet Processes

The paper introduces a novel method using superadditivity to bound the cumulant generating function (CGF) of a Dirichlet Process (DP). This technique converts known asymptotic large deviation principles into practical, non-asymptotic concentration inequalities for sums of independent DPs, providing an upper bound on the CGF.

Open the full episode

More from AI Radio