Online Drift Detection with Maximum Concept Discrepancy

TL;DR

This paper introduces MCD-DD, an innovative, label-free method for detecting concept drift in data streams using maximum concept discrepancy, effective in high-dimensional, real-world scenarios.

Contribution

The paper presents a novel concept drift detection approach based on maximum mean discrepancy and contrastive learning, addressing limitations of existing methods in complex data streams.

Findings

Outperforms existing methods in synthetic and real-world tests

Effective in high-dimensional, irregular distribution shifts

Provides high explainability in drift detection

Abstract

Continuous learning from an immense volume of data streams becomes exceptionally critical in the internet era. However, data streams often do not conform to the same distribution over time, leading to a phenomenon called concept drift. Since a fixed static model is unreliable for inferring concept-drifted data streams, establishing an adaptive mechanism for detecting concept drift is crucial. Current methods for concept drift detection primarily assume that the labels or error rates of downstream models are given and/or underlying statistical properties exist in data streams. These approaches, however, struggle to address high-dimensional data streams with intricate irregular distribution shifts, which are more prevalent in real-world scenarios. In this paper, we propose MCD-DD, a novel concept drift detection method based on maximum concept discrepancy, inspired by the maximum mean discrepancy. Our method can adaptively identify varying forms of concept drift by contrastive learning of concept embeddings without relying on labels or statistical properties. With thorough experiments under synthetic and real-world scenarios, we demonstrate that the proposed method outperforms existing baselines in identifying concept drifts and enables qualitative analysis with high explainability.