Trees to Flows and Back: Unifying Decision Trees and Diffusion Models

▤ Summary

This paper unifies decision trees and diffusion models through a mathematical correspondence between hierarchical trees and diffusion processes. It introduces a shared optimization principle called Global Trajectory Score Matching. The authors show practical value with reeflow, which improves tabular data generation quality while speeding computation, and dsmtree, which distills decision logic into neural networks. The work is research-focused and highlights new methods that could influence future AI tooling and model design.

▦ Linked Companies