Adaptive Monte Carlo via Bandit Allocation

05/13/2014
by   James Neufeld, et al.
0

We consider the problem of sequentially choosing between a set of unbiased Monte Carlo estimators to minimize the mean-squared-error (MSE) of a final combined estimate. By reducing this task to a stochastic multi-armed bandit problem, we show that well developed allocation strategies can be used to achieve an MSE that approaches that of the best estimator chosen in retrospect. We then extend these developments to a scenario where alternative estimators have different, possibly stochastic costs. The outcome is a new set of adaptive Monte Carlo strategies that provide stronger guarantees than previous approaches while offering practical advantages.

READ FULL TEXT

page 6

page 7

page 8

research
10/17/2022

Asymptotic control of the mean-squared error for Monte Carlo sensitivity index estimators in stochastic models

In global sensitivity analysis for stochastic models, the Sobol' sensiti...
research
02/08/2019

Correlated bandits or: How to minimize mean-squared error online

While the objective in traditional multi-armed bandit problems is to fin...
research
02/08/2021

Monte Carlo Rollout Policy for Recommendation Systems with Dynamic User Behavior

We model online recommendation systems using the hidden Markov multi-sta...
research
11/28/2017

More on the restricted almost unbiased Liu-estimator in Logistic regression

To address the problem of multicollinearity in the logistic regression m...
research
09/25/2018

Mostly Harmless Simulations? On the Internal Validity of Empirical Monte Carlo Studies

Currently there is little practical advice on which treatment effect est...
research
10/18/2020

Creative Telescoping on Multiple Sums

We showcase a collection of practical strategies to deal with a problem ...
research
01/02/2022

Global convergence of optimized adaptive importance samplers

We analyze the optimized adaptive importance sampler (OAIS) for performi...

Please sign up or login with your details

Forgot password? Click here to reset