CMStatistics 2018: Start Registration
View Submission - CMStatistics
Title: A minimax near-optimal algorithm for adaptive rejection sampling Authors:  Joseph Lam-Weil - Magdeburg University (Germany) [presenting]
Alexandra Carpentier - Magdeburg University (Germany)
Gilles Blanchard - University of Potsdam (Germany)
Juliette Achdou - Telecom ParisTech (France)
Abstract: Rejection sampling is a fundamental Monte-Carlo method. It is used to sample from distributions admitting a probability density function which can be evaluated exactly at any given point, albeit at a high computational cost. However, without proper tuning, this technique implies a high rejection rate. Several methods have been explored to cope with this problem, based on the principle of adaptively estimating the density by a simpler function, using the information of the previous samples. Most of them either rely on strong assumptions on the form of the density, or do not offer any theoretical performance guarantee. We give the first theoretical lower bound for the problem of adaptive rejection sampling and introduce a new algorithm which guarantees a near-optimal rejection rate in a minimax sense.