Weighted Average Importance Sampling and Defensive Mixture Distributions

By Tim Hesterberg. 1995

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Recommendations as Treatments: Debiasing Learning and Evaluation (2016, neutral): “The SNIPS estimator often has lower variance than the IPS estimator but has a small bias (Hesterberg, 1995).”
Dual Transfer Learning for Neural Machine Translation with Marginal Distribution Regularization (2018, methodology): “This procedure is exactly the technique of importance sampling (Cochran 1977; Hesterberg 1988; 1995).”
Quasi Importance Sampling (2005, methodology): “Then we have to replace this estimate by the ratio estimate [21]”
Controlled stratification for quantile estimation (2008, background): “Other alternatives can be found in Hesterberg (1995).”
Improved Sampling‐Importance Resampling and Reduced Bias Importance Sampling (2003, methodology): “Here, the normalizing constants of h(x) were known, so we could apply the methods in Hesterberg (1995).”
Importance Sampling Squared for Bayesian Inference in Latent Variable Models (2013, methodology): “For the MIXL model, we combine the efficient importance sampling (EIS) method of Richard and Zhang (2007) with the defensive sampling approach of Hesterberg (1995). The importance density is the two component defensive mixture h(xi|yi1,.”
Bayesian Model Comparison for Time-Varying Parameter VARs with Stochastic Volatility (2018, background): “One way to ensure that this finite-variance condition holds is to modify the importance sampling estimator g(h) = g(h; y,𝚺𝜃,𝚺h,𝜽0,h0) to include an additional mixture component as proposed by Hesterberg (1995).”
Bayesian Modelling and Inference on Mixtures of Distributions (2005, background): “For instance, the qit’s can include large tail proposals as in the defensive sampling strategy of Hesterberg (1998), to ensure finite variance. Similarly, Warnes’ (2001) nonparametric Gaussian kernel approximation can be used as a proposal. The generality in the choice of the proposal distributions ”
On the Asymptotic Efficiency of Approximate Bayesian Computation Estimators (2015, background): “The mixture with π(θ) here is to control the importance weight in the tail area (Hesterberg, 1995).”
The Self-Normalized Estimator for Counterfactual Learning (2015, background): “Hence R̂ (h) is equivariant, unlike R̂(h) [19].”
On a Likelihood Approach for Monte Carlo Integration (2004, methodology): “Special cases are Hesterberg’s (1995) stratified importance sampling estimator and Veach and Guibas’s (1995) multiple importance sampling estimator.”