Trees to Flows and Back: Unifying Decision Trees and Diffusion Models (arxiv.org) AI

The arXiv paper “Trees to Flows and Back: Unifying Decision Trees and Diffusion Models” proposes a mathematical correspondence between hierarchical decision trees and diffusion processes, linking them through a shared optimization principle called Global Trajectory Score Matching (GTSM). It argues that (idealized) gradient boosting is asymptotically optimal under this view, and demonstrates practical outcomes via a tabular-generation method (treeflow) and a distillation approach (dsmtree) that transfers tree logic to neural networks with reported near-teacher performance on benchmarks.