An Efficient Hashing-based Ensemble Method for Collaborative Outlier Detection

TL;DR

This paper introduces LSH iTables, an efficient hashing-based ensemble method for collaborative outlier detection that preserves privacy and outperforms existing methods on various datasets.

Contribution

The paper proposes a novel LSH iTables ensemble method that is simple, mergeable, and compatible with differential privacy, improving collaborative outlier detection.

Findings

LSH iTables outperforms recent ensemble methods on multiple datasets.

The method is effective in both centralized and decentralized scenarios.

It maintains high detection accuracy while preserving data privacy.

Abstract

In collaborative outlier detection, multiple participants exchange their local detectors trained on decentralized devices without exchanging their own data. A key problem of collaborative outlier detection is efficiently aggregating multiple local detectors to form a global detector without breaching the privacy of participants' data and degrading the detection accuracy. We study locality-sensitive hashing-based ensemble methods to detect collaborative outliers since they are mergeable and compatible with differentially private mechanisms. Our proposed LSH iTables is simple and outperforms recent ensemble competitors on centralized and decentralized scenarios over many real-world data sets.