Comparison of Long-Term Adaptivity for Neural Networks
Neural Networks can be used for the prognosis of important quality measures in industrial processes to complement or reduce costly laboratory analysis. Problems occur if the system dynamics change over time (concept drift). We survey different approaches to handle concept drift and to ensure good prognosis quality over long time ranges. Two main approaches - data accumulation and ensemble learning - are explained and implemented. We compare the concepts on artificial datasets and on industrial data from three cement production plants and analyse strengths and weaknesses of different approaches.
KeywordsNeural Network Concept Drift Incremental Learning Long Term Learning
Unable to display preview. Download preview PDF.
- 1.Agachi, P.S., Nagy, Z.K., Cristea, M.V., Imre-Lucaci, A.: Model based control: Case Studies in Process Engineering. Wiley-VCH (2006)Google Scholar
- 2.Tsymbal, A.: The problem of concept drift: Definitions and related work. Technical Report, Department of Computer Science, Trinity College: Dublin, Ireland (2004)Google Scholar
- 3.Alsop, P.A.: The Cement Plant Operations Handbook, 5th edn. Tradeship Publications Ltd. (2007)Google Scholar
- 5.Widmer, G., Kubat, M.: Learning in the presence of concept drift and hidden context. Machine Learning 23(1), 69–101 (1996)Google Scholar
- 6.Pérez-Sánchez, B., Fontenla-Romero, O., Guijarro-Berdiñas, B.: An incremental learning method for neural networks in adaptive environments. In: Int. Joint Conf. on Neural Networks (IJCNN 2010), pp. 1–8 (2010)Google Scholar
- 8.Kolter, J.Z., Maloof, M.A.: Dynamic weighted majority: A new ensemble method for tracking concept drift. In: Proc. IEEE Int. Conf. on Data Mining (ICDM 2003), pp. 123–130 (2003)Google Scholar
- 9.He, H.: Self-Adaptive Systems for Machine Intelligence. John Wiley & Sons (2011)Google Scholar