Para receber por e-mail novas publicações de Seminário de Probabilidade e Combinatória, Clique Aqui!

Seminários do IMPA

Probabilidade e Combinatória

Título
Bayesian learning in high-dimensional state-space models
Expositor
Hedibert Lopes

INSPER, São Paulo
Data
Quarta-feira, 17 de abril de 2019, 15:30
Local
Sala 228
Resumo

Applied Bayesian Statistics has benefited greatly from the avalanche of Monte Carlo-based tools for approximate posterior inference in highly complex and structured scientific models. From environmental and health studies to financial applications, virtually all areas of science where evidence-based scrutiny is mandatory to validate scientific hypotheses have benefitted from such technological explosion.

My research also exemplifies these trends. I will discuss work on high-dimensional state-space models with particular attention to time-varying covariance learning. In one direction, my co-authors and I deal with the curse of dimensionality via parameter reduction, such as those found in the factor modelling literature. In another direction, we heavily regularize the estimation of parameters. I will review the challenges we faced when dealing with such high-dimensional state-space models. Two or three motivating examples from my recent research will be used throughout the talk. I finish with a few directions of my future research in these and related areas.