 |
Center for Nonlinear Studies |
Physical systems with many degrees of freedom can often be effectively described in terms of a small number of macro-variables. Coarse-graining in physics follows from detailed knowledge of the underlying laws but this approach fails for many complex systems. I will introduce an approach for coarse-graining arbitrary high-dimensional data based on a hierarchical decomposition of multivariate information. "Local" interactions are discovered and represented as modules that are then progressively combined to capture global dependencies. For gene expression data, for example, coarse-grained states represent a robust computational phenotype capturing diverse biological processes that are useful for targeting treatments. Other example applications include characterizing human behavior and improved covariance estimation for financial data.
Host: Andey Lokhov