Seminar| Institute of Mathematical Sciences
Abstract: Markov chain Monte Carlo (MCMC) is a standard tool for Bayesian statistical inference. Traditional MCMC methods may suffer from the curse of dimensionality and the local-trap problem for inference in sophisticated models. Sequential Monte Carlo (SMC) methods have emerged as alternatives to MCMC methods for complex model inference. In this talk, I will introduce SMC methods for phylogenetic reconstruction and parameter estimation in differential equation models. Firstly, we introduce a new combinatorial SMC method for Bayesian phylogenetic inference, with a novel and efficient proposal distribution. We also explore combining SMC and Gibbs sampling to jointly estimate the phylogenetic trees and evolutionary parameters of genetic data sets. Secondly, we propose an ``embarrassingly parallel'' method for Bayesian phylogenetics, annealed SMC, based on recent advances in the SMC literature, such as adaptive determination of annealing parameters. Thirdly, we consider parameter estimation for nonlinear differential equation (DE) systems from noisy measurements of dynamic systems. We develop a fully Bayesian framework for non-linear DE systems, and derive an SMC method to sample from multi-modal DE posterior distributions.
Tencent Meeting Room Number: 984-921-354