Generalised Random Forest Space Overview

TL;DR

This paper introduces a generalisation space for Random Forests, viewing them as nested ensembles of modules, enabling the development of new methods by exploring different module configurations.

Contribution

It presents a framework that conceptualizes Random Forests as interchangeable modules, facilitating the creation of novel variants and extensions.

Findings

Framework unifies existing RF generalisations

Enables systematic development of new RF methods

Highlights module properties necessary for RF extensions

Abstract

Assuming a view of the Random Forest as a special case of a nested ensemble of interchangeable modules, we construct a generalisation space allowing one to easily develop novel methods based on this algorithm. We discuss the role and required properties of modules at each level, especially in context of some already proposed RF generalisations.