Arithmetical Binary Decision Tree Traversals

TL;DR

This paper proposes novel arithmetic-based methods for traversing binary decision trees, utilizing matrix representations and maximum inner product search to improve understanding and efficiency.

Contribution

It introduces a new approach to binary decision tree traversal using arithmetic operations and matrix representations, offering fresh insights into decision tree algorithms.

Findings

New matrix-based traversal algorithms

Embedding Boolean tests into binary vectors

Enhanced understanding of decision tree structures

Abstract

This paper introduces a series of methods for traversing binary decision trees using arithmetic operations. We present a suite of binary tree traversal algorithms that leverage novel representation matrices to flatten the full binary tree structure and embed the aggregated internal node Boolean tests into a single binary vector. Our approach, grounded in maximum inner product search, offers new insights into decision tree.