Saltar para:
Comuta visibilidade da coluna esquerda
Você está em: Início > Notícias > DaSSWeb - Scalable Bayesian Networks

DaSSWeb - Scalable Bayesian Networks

November 23rd | 14:30

DaSSWeb, Data Science and Statistics Webinar
Scalable Bayesian Networks
Inês Dutra

Assistant professor at the Department of Computer Science of the Faculty of Sciences of University of Porto

Join us here

Bayesian Networks (BNs) are a very powerful model to represent knowledge in the form of a graph, where nodes represent random variables and edges their conditional dependencies. They have been successfully used in many small-to-medium domains such as user-interfaces, robotics, environmental monitoring, health care, among others. Bayesian inference, structure learning and parameter learning are very costly operations as they have exponential complexity depending on the number of edges and nodes in the graph. Several works in the literature have approached the problem of optimizing these operations, either by relaxing the exact problem or resorting to parallelization.
Very few works have succeeded in handling large datasets, specially when learning good quality structures and parameters to the network. In this work, we mitigate these problems by studying efficient ways of implementing structure and parameter learning that can handle large scale data (large number of variables and instances) and that can search for a larger number of alternative networks.

Inês Dutra is an assistant professor at the Department of Computer Science of the Faculty of Sciences of University of Porto. She is also a member of the CINTESIS research unit. Her main research interests are logic programming, (statistical) relational learning and scalable and interpretable machine learning models.


Recomendar Página Voltar ao Topo
Copyright 1996-2022 © Faculdade de Economia da Universidade do Porto  I Termos e Condições  I Acessibilidade  I Índice A-Z  I Livro de Visitas
Página gerada em: 2022-01-24 às 05:56:41 | Política de Utilização Aceitável | Política de Proteção de Dados Pessoais