Safe and Effective Importance Sampling

By Art B. Owen, Yi Zhou. 2000

487 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Numerical Analysis, Statistics and Probability

How researchers see this paper

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

In Artificial Intelligence

In Numerical Analysis

In Statistics and Probability

Citations in the corpus

Learning from Logged Implicit Exploration Data (2010, methodology): “The purpose of τ is to upper-bound the individual terms in the sum and is similar to previous methods like robust importance sampling [10].”
Efficient high-dimensional importance sampling (2007, background): “Owen and Zhou (2000) discuss various improvements of the IS technique which are well suited for low-dimensional applications.”
Efficient Sequential Monte Carlo With Multiple Proposals and Control Variates (2016, methodology): “With control variates ggg ( x ) / q ααα ( x ), where Owen and Zhou (2000)’s regression approach to estimate μ is where”
Adaptive Multiple Importance Sampling (2009, methodology): “The difference with earlier adaptive importance sampling implementations like Population Monte Carlo is that the importance weights of all simulated values, past as well as present, are recomputed at each iteration, following the technique of the deterministic multiple mixture estimator of Owen and ”
Generalized Multiple Importance Sampling (2015, background): “the knowledge of the normalizing constants, prior information of the proposals, etc [Veach and Guibas, 1995; Hesterberg, 1995; Owen and Zhou, 2000; Tan, 2004; He and Owen, 2014; Elvira et al., 2015a].”
MCMC-Driven Adaptive Multiple Importance Sampling (2015, methodology): “MAMIS draws samples from different proposal densities at each iteration, weighting these samples according to the so-called deterministic mixture approach proposed in [11, 14] for a fixed (i.”
Deep Mean Field Games for Learning Optimal Behavior Policy of Large Populations (2017, methodology): “…∈ D samp from k distributions F 1 ( τ ) , . . . , F k ( τ ) , an unbiased estimator of the partition function Z = exp( R W ( τ )) dτ using multiple importance sampling is ˆ Z := 1 M τ j z j exp( R W ( τ j )) (Owen and Zhou, 2000), where importance weights are z j := − 1 (derivation in Appendix F).”
A Sequential Importance Sampling Algorithm for Generating Random Graphs with Prescribed Degrees (2011, methodology): “Important variations which could be coupled with the present algorithms include Owen’s [60] use of control variates to bound the variance and adaptive importance sampling as in Rubinstein [64].”
The sample size required in importance sampling (2015, background): “Owen and Zhou [50] combine Hesterberg’s idea with control variates to give an attractive, practical approach.”
Importance Nested Sampling and the MultiNest Algorithm (2013, methodology): “Owen and Zhou (2000) give a proof that, under the DMMS scheme, the asymptotic variance of the ﬁnal estimator will not be very much worse than that achievable under ordinary importance sampling from the optimal distribution alone.”
A theory of statistical models for Monte Carlo integration (2003, methodology): “Suppose as an alternative that it is feasible to simulate from any or all of the distributions Pσ, as in Geyer (1994), Hesterberg (1995), Meng and Wong (1996) or Owen and Zhou (2000).”
Survey of sampling-based methods for uncertainty and sensitivity analysis (2006, background): “If large sample sizes are required to provide appropriate coverage of low probability/high consequence subsets of values for x, then importance sampling may be a more effective sampling procedure than either random or Latin hypercube sampling [126–134].”
On a Likelihood Approach for Monte Carlo Integration (2004, methodology): “Generally, bridge sampling or stratified mixture sampling involves simulating observations from multiple distributions (see Geyer 1994; Hesterberg 1995; Meng and Wong 1996; Owen and Zhou 2000).”
bridgesampling: An R Package for Estimating Normalizing Constants (2017, unknown): “This stands in sharp contrast to the importance and the generalized harmonic mean estimator for which unwanted tail behavior produces estimators with very large or even inﬁnite variances (e.g., Owen and Zhou 2000; Fr¨uhwirth–Schnatter 2004; Gronau et al. 2017b).”
Importance sampling: a review (2010, methodology): “This approach is particularly attractive when more than one ( p ( x ) , f ( x )) are of interest, and at least one of the chosen q i ( x ) is expected to lead to an eﬃcient IS for each pair (see Theorem 2 in Owen and Zhou 2000).”
Pareto Smoothed Importance Sampling (2015, methodology): “Peruggia (1997) showed that the importance ratios have an infinite variance when leaving out the 21st data point.”
Two-Stage Importance Sampling With Mixture Proposals (2013, methodology): “Owen and Zhou (2000) and Tan (2004) showed that when π(x) is a linear combination of the component proposals, σ2 Z(α) = 0 for any α.”