Unsupervised concept drift detection for multi-label data streams

Many real-world applications adopt multi-label data streams as the need for algorithms to deal with rapidly changing increases. Changes in distribution, also known concept drift, cause existing classification models lose their effectiveness. To assist classifiers, we propose a novel algorithm called Label Dependency Drift Detector (LD3), an unsupervised drift detector using label dependencies within streams. Our study exploits dynamic temporal between labels influence ranking method, which leverages fusion and uses produced detect drift. LD3 is first detection problem area. In this study, perform extensive evaluation of by comparing it 14 prevalent supervised that adapt area 15 datasets baseline classifier. The results show provides 16.9 56% better predictive performance than comparable detectors on both synthetic