Like many other research fields that have turned towards data-driven discovery, human neuroscience has undergone dramatic changes over the course of the last decade. Through new measurement techniques, large collaborative projects and concerted data collection and data sharing efforts, the field is gaining access to data at scales that have never been possible before. And while these data present tremendous opportunities to understand the brain, they also present new challenges that arise from the deluge of data, and from the difficulties that researchers encounter when they attempt to manage, store, analyze and understand it. In this talk, I will present a series of studies that address these challenges through a multi-faceted data science approach: 1) Using deep learning to automate and scale analysis procedures 2) harnessing citizen scientist input to generate labeled training data for machine learning; and 3) developing machine learning algorithms that leverage known brain structure to make inferences about individual variability. I will also discuss the socio-technical challenges facing the field and how we might address these challenges through open-source software development, new collaborative approaches to data analysis and data sharing, as well as through novel approaches to training. Slides are available at: https://arokem.github.io/2019-12-03-CogPer/#/

Ariel Rokem is a senior data scientist at the University of Washington eScience Institute, and an affiliate of the Institute for Neuroengineering, the Center for Computational Neuroscience, and the Center for Studies in Demography and Ecology at the University of Washington. He received a Ph.D. in neuroscience from UC Berkeley (2010) and additional postdoctoral training in computational neuroimaging at Stanford (2011 – 2015). He leads a research program in neuroinformatics: the development of data science tools, techniques and methods, and their application to the analysis of neural data.