Handling Concept Drift in Time-Series Data: Meta-cognitive Recurrent Recursive-Kernel OS-ELM
This paper proposes a meta-cognitive recurrent multi-step-prediction model called Meta-cognitive Recurrent Recursive Kernel Online Sequential Extreme Learning Machine with a new modified Drift Detector Mechanism (Meta-RRKOS-ELM-DDM). This model combines the strengths of Recurrent Kernel Online Sequential Extreme Learning Machine (RKOS-ELM) with the recursive kernel method and a new meta-cognitive learning strategy. We apply Drift Detector Mechanism to solve concept drift problem. Recursive kernel method successfully replaces the normal kernel method in RKOS-ELM and generates a fixed reservoir with optimised information. The new meta-cognitive learning strategy can reduce the computational complexity. The experimental results show that Meta-RRKOS-ELM-DDM has a superior prediction ability in different predicting horizons than the others.
KeywordsTime series Recursive kernel Recurrent Kernel Adaptive Filter Concept drift Meta-cognitive learning
The authors would like to express special thanks of gratitude to UM Grand Challenge from the University of Malaya under Grant GC003A-14HTM, FRGS grant from MOHE FP069-2015A, and the Thailand Research Fund under grant agreement No. TRG5680090 which support our research.
- 1.Kaur, T., Kumar, S., Segal, R.: Application of artificial neural network for short term wind speed forecasting. In: Proceedings of 2016 Biennial International Conference on Power and Energy Systems: Towards Sustainable Energy, pp. 1–5 (2016)Google Scholar
- 4.Hammoodi, M., Stahl, F., Tennant, M.: Towards online concept drift detection with feature selection for data stream classification. In: Proceedings of the 22nd European Conference on Artificial Intelligence (2016)Google Scholar
- 6.Kulkarni, P., Ade, R.: Logistic regression learning model for handling concept drift with unbalanced data in credit card fraud detection system. In: Satapathy, S.C., Raju, K.S., Mandal, J.K., Bhateja, V. (eds.) Proceedings of the Second International Conference on Computer and Communication Technologies. AISC, vol. 380, pp. 681–689. Springer, New Delhi (2016). https://doi.org/10.1007/978-81-322-2523-2_66CrossRefGoogle Scholar
- 7.Liu, Z., Loo, C.K., Masuyama, N., Pasupa, K.: Multiple steps time series prediction by a novel recurrent kernel extreme learning machine approach. In: Proceeding of the 9th International Conference on Information Technology and Electrical Engineering, p. SIG5.5 (2017)Google Scholar
- 11.Hermans, M.: Expanding the theoretical framework of reservoir computing. Ph.D. thesis, Ghent University (2012)Google Scholar
- 12.Yahoo: Shanghai stock exchange (2017). https://finance.yahoo.com/quote/000001.SS?p=000001.SS
- 13.Ministry of the Environment and Climate Change in Ontario: Air pollutant data (2000). http://www.airqualityontario.com/history