Description: By linking conceptual theories with observed data, transparent hierarchical generative models can support reasoning in complex situations. They have come to play a central role both within and beyond statistics, providing the basis for power analysis in molecular biology, microbiome studies, and resource allocation in epidemiology. These lectures will survey some applications of transparent generative models and show how they inform experimental design, iterative model refinement, goodness-of-fit evaluation, and agent based simulation. We emphasize a modular view of generative mechanisms and discuss how they can be flexibly recombined in new problem contexts. 

Current research in generative models is currently split across several islands of activity, and we highlight opportunities lying at disciplinary intersections. We will study both coded applications that go over these ideas and the applications will be done using R. All attendees will be asked to have a recent version of R installed on their laptop so we can do hands-on activities together.

We will go over the principles of:

- hierarchical mixture models (https://www.huber.embl.de/msmb/)

- latent variables for microbiome (https://doi.org/10.6084/m9.figshare.c.6922510.v1)

- machine learning for goodness of fit

( https://github.com/krisrs1128/generative_review. )

Registration can be completed here.

Joint Seminarof Analysis & MatHBioS Thematic Line

Speaker:  Prof.Fabio Chalub (NOVA Math, Universidade NOVA de Lisboa, Portugal)

Date/time: 07/05/2025 (Wednesday), from 14:15 to 15:15.

Location:  Room 1, Building XI.

Title:  Group centrality in optimal and suboptimalvaccination for epidemic models in contact networks

Abstract:  The pursuitof strategies that minimize the number of individuals needing vaccination tocontrol an outbreak is a well-established area of study in mathematicalepidemiology. However, when vaccines are in short supply, public policy tendsto prioritize immunizing vulnerable individuals over epidemic control. As aresult, optimal vaccination strategies may not always be practical forinforming real-world public policies. In this work, we focus on a disease thatresults in long-term immunity and spreads through a heterogeneous population,represented by a contact network. We study four well-known group centralitymeasures and show that the GED-Walk offers a reliable means of estimating theimpact of vaccinating specific groups of individuals, even in suboptimal cases.Additionally, we depart from the search for target individuals to be vaccinatedand provide proxies for identifying optimal groups for vaccination. While theGED-Walk is the most useful centrality measure for suboptimal cases, thebetweenness (a related, but different centrality measure) stands out whenlooking for optimal groups. This indicates that optimal vaccination is notconcerned with breaking the largest number of transmission routes, butinterrupting geodesic ones.

This is a joint work with Jorge O. Cerdeira andMatheus Hansen (NOVA Math, Universidade NOVA de Lisboa, Portugal)

Description: By linking conceptual theories with observed data, transparent hierarchical generative models can support reasoning in complex situations. They have come to play a central role both within and beyond statistics, providing the basis for power analysis in molecular biology, microbiome studies, and resource allocation in epidemiology. These lectures will survey some applications of transparent generative models and show how they inform experimental design, iterative model refinement, goodness-of-fit evaluation, and agent based simulation. We emphasize a modular view of generative mechanisms and discuss how they can be flexibly recombined in new problem contexts. 

Current research in generative models is currently split across several islands of activity, and we highlight opportunities lying at disciplinary intersections. We will study both coded applications that go over these ideas and the applications will be done using R. All attendees will be asked to have a recent version of R installed on their laptop so we can do hands-on activities together.

We will go over the principles of:

- hierarchical mixture models (https://www.huber.embl.de/msmb/)

- latent variables for microbiome (https://doi.org/10.6084/m9.figshare.c.6922510.v1)

- machine learning for goodness of fit

( https://github.com/krisrs1128/generative_review. )

Registration can be completed here.