Local Adaptive Importance Sampling for Multivariate Densities with Strong Nonlinear Relationships

By Geof H. Givens, Adrian E. Raftery. 1996

How researchers see this paper

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

1. Arnaud Doucet
2. Brooks Paige
3. P. de Valpine
4. Yoshua Bengio
5. Nando de Freitas
6. Ingmar Schuster
7. Dieter Fox
8. Ruslan Salakhutdinov

Citations in the corpus

• Kernel Sequential Monte Carlo (2015, methodology): “ AIS does not use re-sampling steps between iterations and thus produces a particle system that is unconditionally i.i.d. at every iteration. The Local Adaptive Importance Sampling (LAIS) approach by [23] has a similar effect as KASS with a nonlinear kernel regarding adaptation to local covariance st”
• Sequential Monte Carlo Methods in Practice (2001, methodology): “(Givens and Raftery 1996), however, disagree with the use of the same variance matrix BmJ for all kernels at stage j.”
• Monte Carlo State-Space Likelihoods by Weighted Posterior Kernel Density Estimation (2004, methodology): “The MCKL method appears to be a new approach to maximum likelihood parameter estimation for state-space models, but it is related to kernel density estimation of Bayesian posterior densities (West 1993;Chen 1994;Givens and Raftery 1996; Liu and West 2001).”
• Improved Sampling‐Importance Resampling and Reduced Bias Importance Sampling (2003, methodology): “Givens & Raftery (1996) considered adaptive versions of the SIR algorithm, where SIR was rerun several times, and for each new run the proposal distribution was updated using kernel density estimates.”
• Fitting Population Dynamics Models to Count and Cull Data Using Sequential Importance Sampling (2000, methodology): “Global kernel density estimation was also used by Huirzeler and Kiinsch (1995), and Givens and Raftery (1996) developed a method of local kernel density estimation for use when contours of equiprobability are far from elliptical.”
• Combined Parameter and State Estimation in Simulation-Based Filtering (2001, methodology): “For example, it has been extended to include variable shapes of multivariate kernels to re ect changing patterns of dependencies among parameters in di erent regions of parameter space (Givens and Raftery 1996), as explicitly anticipated in West (1993a,b). A related approach may be referred to as th”
• Sequential Monte Carlo samplers (2002, methodology): “Givens and Raftery (1996). However, this approach is problematic.”