Abstract
Mining multi-label data stream is a huge challenge due to its properties of multiple labels and dynamic streaming characteristics, where each label may experience different con-cept drifts simultaneously or distinctly and class imbalance, etc. Existing works either concentrate on only one type of drift for all labels or ignore the dynamic class imbalance, which usually re-sults in a great degeneration in classification performance. To this end, we propose an efficient ensemble based on the Self-Adjusting Memory (SAM) and Enhanced Punishment mechanism, named as SAMEP, where the Self Adjusting Memory is used to adapt to the heterogeneous concept drift for different labels in multi-label data and the Enhanced Punishment mechanism is utilised to deal with the mining difficulties brought by multiple labels, such as class imbalance, etc. Extensive experiments conducted on multi-label datasets for six common metrics demonstrate the effectiveness of our proposed algorithm as compared to 7 latest well-known multi-label data stream classification methods demonstrated the effectiveness of our proposed method. In addition, non-parametric statistical (the Friedman test with Nemenyi post-hoc) analysis validate the effectiveness of our proposed framework.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.