Computational Methods for Bayesian Inference

33 Foro Nacional de Estadistica (FNE)y

13 Congreso Latinoamericano de Sociedades de Estadistica (CLATSE)

Ciudad de Guadalajara

October 3-5, 2018

Hedibert Freitas Lopes

Professor of Statistics and Econometrics

Insper Institute of Education and Research

URL: www.hedibert.org

E-mail: hedibertfl@insper.edu.br

http://hedibert.org/wp-content/uploads/2018/08/ComputationalMethodsforBayesianInference.html

Summary: All three lectures will start with simple, toy examples to introduce concepts and illustrate their implementation and will finish with more serious, complex studies based on my own research over the years.

Lecture 1: Bayesian Inference via Gibbs, Metropolis and other MCMC Schemes

· Basic ingredients

· Monte Carlo methods

· Markov Chain Monte Carlo methods

· Forward filtering, backward sampling

· Paper 1: Parsimony inducing priors for large scale state-space models

Lecture 2: Sequential Monte Carlo for Bayesian online learning

· Sequential Monte Carlo methods: state learning

· Sequential Monte Carlo methods: state and parameter learning

· Paper 2: Particle Learning for Bayesian Semi-Parametric Stochastic Volatility

Lecture 3: Bayesian infereequential Monte Carlo methods

· Paper 3: Efficient sampling for Gaussian linear regression with arbitrary priors

· Paper 4: Efficient Bayesian Inference for Multivariate Factor SV Models