EIFuzzCND: uma estratégia incremental para classificação multiclasse e detecção de novidades em fluxos de dados

Abstract

The study addresses novelty detection in data streams, emphasizing the significance of this task in high-volume and high-velocity information environments. It proposes substantial improvements to the EFuzzCND algorithm, leading to the development of EIFuzzCND. These enhancements encompass an incremental approach, a reduction in dependence on true labels, and the implementation of the Incremental Confusion Matrix. Experiments validate the efficacy of EIFuzzCND across diverse scenarios, and result analysis underscores its capability to handle specific challenges, such as sudden concept shifts. The work contributes to advancing novelty detection in data streams by providing an innovative and practical approach, concluding with recommendations for future research.