Item

Abstract

The study of brain function requires collecting and analyzing highly complex and multivariate datasets. Modern machine learning techniques are useful at several stages of the analysis. First unsupervised learning techniques based on tools such as non-negative matrix factorization helps identify the relevant features and underlying structure of the data. Second, statistical analysis based on kernel embedding of distributions help identify complex interactions between different aspects of neural activity. Finally, causal inference allows estimating the directionality of information transfer across brain networks. During this tutorial, we will implement and use some of these tools to analyze intercortical recordings and explain how they help neuroscientists understand brain function.