Approximate Inference for Generic Likelihoods via Density-Preserving GMM Simplification
Research output: Conference Papers › RGC 32 - Refereed conference paper (without host publication) › peer-review
Author(s)
Related Research Unit(s)
Detail(s)
Original language | English |
---|---|
Publication status | Published - Dec 2016 |
Conference
Title | NIPS 2016 Workshop on Advances in Approximate Bayesian Inference |
---|---|
Location | Barcelona |
City | |
Period | 9 December 2016 |
Link(s)
Permanent Link | https://scholars.cityu.edu.hk/en/publications/publication(25e542b6-6274-4b14-b34a-49c4bac82223).html |
---|
Abstract
We consider recursive Bayesian filtering where the posterior is represented as a
Gaussian mixture model (GMM), and the likelihood function as a sum of scaled
Gaussians (SSG). In each iteration of filtering, the number of components increases.
We propose an algorithm for simplifying a GMM into a reduced mixture model
with fewer components, which is based on maximizing a variational lower bound
of the expected log-likelihood of a set of virtual samples. We also propose an
efficient algorithm for approximating an arbitrary likelihood function as an SSG.
Experiments on synthetic 2D GMMs, simulated belief propagation and visual
tracking show that our algorithm can be widely used for approximate inference.
Gaussian mixture model (GMM), and the likelihood function as a sum of scaled
Gaussians (SSG). In each iteration of filtering, the number of components increases.
We propose an algorithm for simplifying a GMM into a reduced mixture model
with fewer components, which is based on maximizing a variational lower bound
of the expected log-likelihood of a set of virtual samples. We also propose an
efficient algorithm for approximating an arbitrary likelihood function as an SSG.
Experiments on synthetic 2D GMMs, simulated belief propagation and visual
tracking show that our algorithm can be widely used for approximate inference.
Citation Format(s)
Approximate Inference for Generic Likelihoods via Density-Preserving GMM Simplification. / YU, Lei; YANG, Tianyu; Chan, Antoni B.
2016. Paper presented at NIPS 2016 Workshop on Advances in Approximate Bayesian Inference.
2016. Paper presented at NIPS 2016 Workshop on Advances in Approximate Bayesian Inference.
Research output: Conference Papers › RGC 32 - Refereed conference paper (without host publication) › peer-review