Enhash: A Fast Streaming Algorithm For Concept Drift Detection
This work addresses the challenge of efficient concept drift detection for streaming data applications, but it appears incremental as it builds on existing ensemble learners with a focus on speed.
The paper tackles the problem of detecting concept drift in data streams by proposing Enhash, a fast ensemble learner that achieves competitive performance to existing methods in significantly less time, with moderate resource requirements, as demonstrated on 6 artificial and 4 real datasets.
We propose Enhash, a fast ensemble learner that detects \textit{concept drift} in a data stream. A stream may consist of abrupt, gradual, virtual, or recurring events, or a mixture of various types of drift. Enhash employs projection hash to insert an incoming sample. We show empirically that the proposed method has competitive performance to existing ensemble learners in much lesser time. Also, Enhash has moderate resource requirements. Experiments relevant to performance comparison were performed on 6 artificial and 4 real data sets consisting of various types of drifts.