MLAB

Careful methods for problems that matter

MLAB is a research lab at the Institute of Informatics of the Federal University of Rio Grande do Sul (UFRGS), based in Porto Alegre, Brazil. We bring together researchers and students who share an interest in machine learning for biomedical data and in doing that work with the rigor and care it deserves.

Our research runs along two threads. On the applied side, we work with real biomedical data clinical records, medical images, genomic sequences, population-level health data with all the complexity and imperfection that come with it. On the foundational side, we develop and investigate machine learning methods and solutions: designing new approaches, improving existing ones, evaluating them rigorously, reducing bias, and understanding whether experimental evidence supports their intended use. We do both because both matter and because, in our experience, each keeps the other from becoming fragile.

Decisions made with the help of a model can affect real people, and often affect them unevenly. Making machine learning more reliable, less biased, and safer to use is part of our research. We seek to understand how the quality and reliability of these solutions can be improved, and how rigorous evidence can help determine when and where they may be useful. Ultimately, we hope our work can contribute to better outcomes when machine learning is applied to health for clinicians, for patients, and for the systems that support them.