The fortnightly BRAG meeting will be held this Thursday, August 22nd, at 1 pm via Zoom/GP-Y801. This week we will have a presentation by Matthew Adams and Harriette Phillips along with discussion.
Zoom link: https://qut.zoom.us/j/85019677992?pwd=mDRlg0MIu8bxRI3mutK5RmeypJaGuu.1
Meeting ID: 850 1967 7992
Passcode: 130511
Matthew’s Talk
Title: Moment matching for truncated multivariate Gaussian distributions
Abstract: Calculating the moments of truncated multivariate Gaussian distributions from the moments of the associated non-truncated distributions is an already resolved problem. The inverse problem is more challenging: calculating the moments of a non-truncated multivariate Gaussian distribution if we know information about the moments of the distribution truncated within some shape (such as a hyper-rectangle or a simplex). This inverse problem is called “moment matching”. In this talk I discuss some preliminary work on moment matching for truncated multivariate Gaussian distributions, using a recently proposed approach that transforms the problem into a system of ordinary differential equations to solve numerically. I could imagine that this problem, if fully resolved, might be particularly helpful for dealing with Bayesian inference applications where data cannot exit some finite bounds (such as non-negative abundances of animals, or fluid compositions consisting of non-negative fractional components that must sum to one).
Harriette’s Talk
Title: Vulnerability Indices for Children’s Health and Wellbeing
Abstract: Child health is an important factor of overall community health. Disparities in children’s health can affect adult health and wellbeing, economic success, and education ultimately leading to a huge cost to health care systems. Children who have not had their needs met are more vulnerable to harm. Furthermore, differences across regions can influence vulnerability where regions with better access to services and resources are likely to have better outcomes. Recently there has been an increased interest in vulnerability research as it allows for the prioritisation of policies, actions, and decision-making. However, vulnerability cannot be directly observed. Instead, it is measured through characteristics of vulnerability and how they differ between regions. Often this is achieved through the development of vulnerability indices. This project aims to develop vulnerability indices for children in Australia using data from the Australian Child and Youth Wellbeing Atlas.