Recent Research Projects

I am looking for undergraduate research students and MSc students to join PPGCC-CM/UTFPR. If you are interested, please review my recent research projects and feel free to contact me via email. I am also open to supervising students on other related research topics.

• Investigation of Open Data from Brazilian Higher Education

  In recent years, there has been a growing trend toward the release of open data across various government sectors. In particular, the Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira (INEP) makes a wealth of data related to higher education publicly available each year through the Higher Education Census and ENADE. However, these datasets are still underexplored by both the academic community and governmental bodies. Rarely are they analyzed alongside other relevant information, such as demographic and economic development data. This project aims to employ data mining techniques to uncover patterns and insights that can provide valuable knowledge about Brazilian higher education for society. The goal is to offer empirical evidence on a range of issues related to public educational policies, including the expansion and regulation of course offerings, the challenge of student dropout rates, and the identification of key quality and performance indicators, among others.

• Machine Learning Techniques to Support Herbarium Specimen Identification

  Herbariums play a vital role in cataloging plant specimens collected from nature, preserving them as exsiccates -— dried samples mounted on boards with detailed descriptions. These specimens are essential for taxonomic, biogeographic, and ecological studies, but their identification is crucial for accuracy. However, identification remains a bottleneck in herbarium workflows due to its complexity, susceptibility to errors, and heavy reliance on the expertise and availability of specialists. To address this challenge, recent initiatives have utilized images from virtual herbariums to create and share datasets. These datasets are used to train machine learning models aimed at supporting specialists in the identification process. However, most studies have focused on databases created for competitions, such as Herbarium2019 and Herbarium 2021 Half-Earth, or on collections from herbariums in Europe, the Americas, and Asia. As of now, no dedicated database for Brazilian plant species has been proposed or evaluated. In this context, the present project proposes the application of machine learning techniques to classify herbarium specimens of endemic Brazilian species. The goal is to develop tools that assist herbariums in automating the identification process. This initiative aims to fill a significant gap, particularly given that many Brazilian species are ecologically important and are often threatened with extinction due to habitat loss.