Weighted Ensemble with Dynamical Chunk Size for Imbalanced Data Streams in Nonstationary Environment

DOI

10.2991/iccia-17.2017.60How to use a DOI?

Keywords

Imbalanced data, concept drift, dynamic chunk, weighted ensemble

Abstract

In recent years, learning from data stream has been more and more popular because of its extensive applications. However, most algorithms assume there are no concept drift in one chunk, as the performance of evaluation is sensitive to the chunk size. In this paper, we propose a new approach (WEDC) by introducing the concept drift detection mechanism to dynamically adjusting the chunk size. In addition, we add weighted mechanism to ensemble classifiers, which make WEDC could react to different types of concept drifts well. Experiments performed on the synthetic datasets show that our approach is competitive in the predication accuracy for data streams including different kinds of concept drifts.

Copyright

© 2017, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Nini Liu
AU  - Wen Zhu
AU  - Bo Liao
AU  - Siqi Ren
PY  - 2016/07
DA  - 2016/07
TI  - Weighted Ensemble with Dynamical Chunk Size for Imbalanced Data Streams in Nonstationary Environment
BT  - Proceedings of the 2nd International Conference on Computer Engineering, Information Science & Application Technology (ICCIA 2017)
PB  - Atlantis Press
SP  - 352
EP  - 355
SN  - 2352-538X
UR  - https://doi.org/10.2991/iccia-17.2017.60
DO  - 10.2991/iccia-17.2017.60
ID  - Liu2016/07
ER  -