THAAD: Efficient Matching Queries under Temporal Abstraction for Anomaly Detection

TL;DR

This paper introduces a novel algorithm and data structure for anomaly detection in temporal data, utilizing symbolic time intervals and gradient abstraction to efficiently identify unusual patterns with superior performance.

Contribution

The paper proposes a new approach combining gradient temporal abstraction with a dynamic data structure for efficient anomaly detection in time-series data.

Findings

Outperforms baseline methods on DNS network traffic data

Supports polylogarithmic update and query times

Introduces a new parameter for pairwise symbol difference control

Abstract

In this paper we present a novel algorithm and efficient data structure for anomaly detection based on temporal data. Time-series data are represented by a sequence of symbolic time intervals, describing increasing and decreasing trends, in a compact way using gradient temporal abstraction technique. Then we identify unusual subsequences in the resulting sequence using dynamic data structure based on the geometric observations supporting polylogarithmic update and query times. Moreover, we introduce a new parameter to control the pairwise difference between the corresponding symbols in addition to a distance metric between the subsequences. Experimental results on a public DNS network traffic dataset show the superiority of our approach compared to the baselines.