Evento ocorre no dia 18 de novembro, das 10h30 às 12h, no Anfiteatro do CIn
O Centro de Informática (CIn) da UFPE recebe a Profa. Maria Rosário Oliveira, do Departamento de Matemática do Instituto Técnico da Universidade de Lisboa, para apresentação de Seminário de Pesquisa intitulado “From Modeling to Insight: Fisher’s Discriminant Analysis and Visualization Tools for Interval Data”. O evento ocorre no dia 18 de novembro, das 10h30 às 12h, no Anfiteatro do CIn. A atividade é gratuita e aberta a comunidade do CIn-UFPE. Não é necessário realizar inscrição.
Confira abaixo mais informações sobre o Seminário de Pesquisa:
Palestrante: M. Rosário Oliveira, CEMAT & Dep. Mathematics, Instituto Superior Técnico, Lisbon, Portugal
Título: From Modeling to Insight: Fisher’s Discriminant Analysis and Visualization Tools for Interval Data
Resumo:
In Data Science, entities are typically described by single-valued measurements, yet many applications involve data with inherent internal variability. Symbolic Data Analysis provides a framework to model such complex structures, including interval-valued and histogram-valued data. In this work, we extend the conventional multiclass Fisher’s Discriminant Analysis to interval-valued data, using Moore’s algebraic structure for interval operations and the Mallows (L2 Wasserstein) distance. Within this framework, Fisher’s objective function is reformulated to separately account for the contributions of the interval’s centers and ranges, and it is numerically optimized to obtain discriminant functions for classification. Classifier performance is evaluated through both class-specific and global measures, which typically assume error-free labels. To complement these measures, we employ visualization tools based on the Mallows’ distance, improving interpretability and revealing the influence of outliers and classification uncertainty. The proposed approach is illustrated with real data, demonstrating its effectiveness and the advantages of the visual diagnostics.
Joint work with Diogo Pinheiro (Dep. Mathematics, Instituto Superior Técnico, Lisbon, Portugal) and Lina Oliveira (CAMGSD & Dep. Mathematics, Instituto Superior Técnico, Lisbon, Portugal).
Mini-bio: M. Rosário Oliveira holds a Ph.D. in Mathematics and Habilitation in Mathematics from Instituto Superior Técnico (IST), University of Lisbon, where she is currently an Associate Professor in the Department of Mathematics and a researcher at the Center for Computational and Stochastic Mathematics (CEMAT). Her doctoral work focused on robust statistics, and her research interests include data science, robust multivariate analysis, and symbolic data analysis. She has been involved in several national and international research projects in these areas.
Comentários desativados