Data Streams Fusion by Frequent Correlations Mining
Applications acquiring data from multiple sensors have to properly refer data to observables. On-line classification and clustering as basic tools for performing information fusion are computationally viable. However, they poorly exploit temporal relationships in data as patterns mining methods can do. Hence, this paper introduces a new algorithm for the correlations mining in the proposed graph-stream data structure. It can iteratively find relationships in complex data, even if they are partially unsynchronized or disturbed. Retrieved patterns (traces) can be used directly to fuse multi-perspective observations. The algorithm’s evaluation was conducted during experiments on artificial data sets while its computational efficiency and results quality were measured.
KeywordsInformation fusion Correlations mining Complex data
- 1.Fricker, D., Zhang, H., Yu, C.: Sequential pattern mining of multimodal data streams in dyadic interactions. In: 2011 IEEE International Conference on Development and Learning (ICDL), vol. 2, pp. 1–6 (2011)Google Scholar
- 4.Matsubara, Y., Sakurai, Y., Faloutsos, C.: Autoplait: automatic mining of co-evolving time sequences. In: Proceedings of the 2014 ACM SIGMOD International Conference on Management of Data, SIGMOD 2014, pp. 193–204. ACM, New York (2014)Google Scholar
- 6.Nguyen, H.L., Woon, Y.K., Ng, W.K.: A survey on data stream clustering and classification. Knowl. Inf. Syst. 1–35 (2014)Google Scholar
- 7.Oates, T., Cohen, P.R.: Searching for structure in multiple streams of data. In: Proceedings of the Thirteenth International Conference on Machine Learning, pp. 346–354. Morgan Kaufmann (1996)Google Scholar
- 9.Wang, Z., Qian, J., Zhou, M., Dong, Y., Chen, H.: Mining group correlations over data streams. In: 2011 6th International Conference on Computer Science Education (ICCSE), pp. 955–959, August 2011Google Scholar