Machine learning and computational cognitive neuroscience

21 September 2018
By Francisco Pereira  (NIMH, Maryland, USA)
Time: 16:00 / Location: Amphitheater 2 FC6 (Dep. Ciência de Computadores-FCUP)
Abstract: Across scientific fields, researchers increasingly require advanced computational methods to do their work. The applications range from processing large datasets to transforming them into meaningful quantitative measures, testing hypotheses, or even helping researchers formulate new models of the underlying natural phenomena. This is especially true in cognitive neuroscience, where much of the work is based on complex brain imaging datasets. In this talk, I will review brain imaging technology, and highlight some of the more interesting uses of machine learning to further cognitive neuroscience research. I will also describe several areas where there are opportunities to make methodological contributions with clinical and scientific impact. If time permits, I will introduce our own work towards building universal decoders of mental representations from brain images.
Short Bio: Francisco Pereira leads the (recently created) Machine Learning Team at the National Institute of Mental Health, in Bethesda, Maryland. The mission of the team is to assist researchers who want to use machine learning methods to analyze brain imaging and behavioral data, and also to develop new analysis approaches. Prior to that, he was a staff scientist at Siemens Healthcare, where he managed the Computational Neuroscience program. He did his postdoc at the Princeton Neuroscience Institute, working with Matt Botvinick and Ken Norman, as well as anyone stopping by his office with an interesting question and chocolate. He has a Ph.D. in Computer Science and Neural Basis of Cognition from Carnegie Mellon University, where he worked with Tom Mitchell and Marcel Just, and an undergraduate degree in Computer Science from Universidade do Porto. He promises not to talk about himself in the third person during this talk.