Extending Decision Predicate Graphs for Comprehensive Explanation of Isolation Forest

Matteo Ceschin, Leonardo Arrighi, Luca Longo, Sylvio Barbon Junior · 2025

Read Extending Decision Predicate Graphs for Comprehensive Explanation of Isolation Forest by Matteo Ceschin, Leonardo Arrighi, Luca Longo, Sylvio Barbon Junior (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-04019.txt.

Description

Matteo Ceschin, Leonardo Arrighi, Luca Longo, and Sylvio Barbon Junior present a way to explain how an outlier detector reaches its verdicts, in this 2025 paper. The detector, Isolation Forest, flags odd samples by isolating them in many random trees, and it works well. The more trees it grows, though, the harder its choices are to explain, and preprocessing steps can hide bias too.

The authors extend a tool called the Decision Predicate Graph, which lays out the logic of an ensemble as a graph of conditions. Their version adds a score that follows how each sample's status as inlier or outlier spreads through the graph. The result shows which features drive outlier calls and where the decision boundary sits for ordinary points. The team presents the method as a step toward pipelines whose every stage, not just the final model, can be inspected and trusted.

Catalog record

CREATORMatteo Ceschin, Leonardo Arrighi, Luca Longo, Sylvio Barbon Junior
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-04019

Browse the full library · About the Tower of Babel