ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 1006.3002
107
8

Free energy Sequential Monte Carlo, application to mixture modelling

15 June 2010
Nicolas Chopin
Pierre E. Jacob
ArXivPDFHTML
Abstract

We introduce a new class of Sequential Monte Carlo (SMC) methods, which we call free energy SMC. This class is inspired by free energy methods, which originate from Physics, and where one samples from a biased distribution such that a given function ξ(θ)\xi(\theta)ξ(θ) of the state θ\thetaθ is forced to be uniformly distributed over a given interval. From an initial sequence of distributions (πt)(\pi_t)(πt​) of interest, and a particular choice of ξ(θ)\xi(\theta)ξ(θ), a free energy SMC sampler computes sequentially a sequence of biased distributions (π~t)(\tilde{\pi}_{t})(π~t​) with the following properties: (a) the marginal distribution of ξ(θ)\xi(\theta)ξ(θ) with respect to π~t\tilde{\pi}_{t}π~t​ is approximatively uniform over a specified interval, and (b) π~t\tilde{\pi}_{t}π~t​ and πt\pi_{t}πt​ have the same conditional distribution with respect to ξ\xiξ. We apply our methodology to mixture posterior distributions, which are highly multimodal. In the mixture context, forcing certain hyper-parameters to higher values greatly faciliates mode swapping, and makes it possible to recover a symetric output. We illustrate our approach with univariate and bivariate Gaussian mixtures and two real-world datasets.

View on arXiv
Comments on this paper