Title: Tree-Structured Boosting: Intrinsic Connections Between CART & Gradient Boosting

Speaker:

Gilmer Valdes, Assistant Clinical Professor, Radiation Oncology, School of Medicine,UCSF

Abstract:

Additive models, such as produced by gradient boosting, & full order interaction models, such as classification & regression trees (CART), are widely used algorithms that have been investigated largely in isolation. We show that these models exist along a spectrum, identifying deep connections between these two approaches. We introduce a novel technique called tree-structured boosting for creating Decision Trees, & shows that this method can produce models equivalent to CART or gradient boosting at the extremes by varying a single parameter. Notably, tree-structured boosting can produce hybrid models between CART & gradient boosting that can outperform either of these approaches