Anomaly Detection in Audio with Concept Drift using Adaptive Huffman Coding

TL;DR

This paper introduces an adaptive Huffman coding approach for anomaly detection in audio data that accounts for concept drift, outperforming existing methods like AGMM in accuracy and efficiency.

Contribution

It presents a novel adaptive Huffman coding method that dynamically adjusts to concept drift without prior cluster information, improving anomaly detection in audio.

Findings

Higher AUC compared to AGMM and fixed Huffman trees

More time-efficient than existing methods

Effective in handling concept drift in audio data

Abstract

When detecting anomalies in audio, it can often be necessary to consider concept drift: the distribution of the data may drift over time because of dynamically changing environments, and anomalies may become normal as time elapses. We propose to use adaptive Huffman coding for anomaly detection in audio with concept drift. Compared with the existing method of adaptive Gaussian mixture modeling (AGMM), adaptive Huffman coding does not require a priori information about the clusters and can adjust the number of clusters dynamically depending on the amount of variation in the audio. To control the size of the Huffman tree, we propose to merge clusters that are close to each other instead of replacing rare clusters with new data. This reduces redundancy in the Huffman tree while ensuring that it never forgets past information. On a dataset of audio with concept drift which we have curated ourselves, our proposed method achieves higher area under the curve (AUC) compared with AGMM and fixed-length Huffman trees. The proposed approach is also time-efficient and can be easily extended to other types of time series data (e.g., video).