Density Estimation Trees as fast non-parametric modelling tools

TL;DR

Density Estimation Trees (DETs) are fast, scalable non-parametric models for probability density estimation, suitable for resource-intensive applications like LHC data analysis, despite being less accurate than kernel methods.

Contribution

This paper introduces DETs as efficient, parallelizable density estimation tools and provides an implementation within RooFit for high-energy physics applications.

Findings

DETs are computationally fast and scalable.

DETs are suitable for resource-intensive data analysis tasks.

The paper provides a RooFit implementation for the HEP community.

Abstract

Density Estimation Trees (DETs) are decision trees trained on a multivariate dataset to estimate its probability density function. While not competitive with kernel techniques in terms of accuracy, they are incredibly fast, embarrassingly parallel and relatively small when stored to disk. These properties make DETs appealing in the resource-expensive horizon of the LHC data analysis. Possible applications may include selection optimization, fast simulation and fast detector calibration. In this contribution I describe the algorithm, made available to the HEP community in a RooFit implementation. A set of applications under discussion within the LHCb Collaboration are also briefly illustrated.