Neighborhood density estimation using space-partitioning based hashing schemes

TL;DR

This paper presents FiRE/FiRE.1, a new sketching algorithm for anomaly detection in large-scale single-cell RNA data, and Enhash, an efficient ensemble learner for concept drift detection in streaming data, both outperforming existing methods.

Contribution

Introduces FiRE/FiRE.1 for rapid anomaly detection in single-cell data and Enhash for efficient concept drift detection, advancing current techniques.

Findings

FiRE/FiRE.1 outperforms state-of-the-art anomaly detection methods.

Enhash demonstrates high accuracy and speed in detecting concept drift.

Both methods are resource-efficient and suitable for large-scale data.

Abstract

This work introduces FiRE/FiRE.1, a novel sketching-based algorithm for anomaly detection to quickly identify rare cell sub-populations in large-scale single-cell RNA sequencing data. This method demonstrated superior performance against state-of-the-art techniques. Furthermore, the thesis proposes Enhash, a fast and resource-efficient ensemble learner that uses projection hashing to detect concept drift in streaming data, proving highly competitive in time and accuracy across various drift types.