Knowledge Matching in Horizontal Collaborative Fuzzy Clustering
In horizontal collaborative fuzzy clustering, the collaboration is implemented by adapting the difference between the collaborated partition matrix and the collaborating partition matrixes (called knowledge) in the objective function. Given a partition matrix of a dataset, the matrix obtained by exchanging its rows arbitrarily is also a partition matrix of the dataset. These partition matrices are the same except the orders of the rows, thus they represent the same knowledge. Let the original partition matrix be the collaborated one, the other be the collaborating ones, the final partition matrix given by collaborative fuzzy clustering should be the same as the original one. But the experiments result is different. If we change the collaborating matrixes to the same as the collaborated one, the final partition matrix remains the same as the collaborated one. This tells us the matching of the rows of the collaborated matrix and the collaborating matrixes is necessary. Furthermore, similar problems happened in practice. To deal with this problem, this paper proposes a knowledge matching algorithm based on the Kuhn Munkras (KM) algorithm in bipartite graph matching, then gives an improved horizontal collaborative fuzzy clustering algorithm. Experimental results show that the proposed horizontal collaborative clustering algorithm has superior performance than the existing ones.
KeywordsFuzzy clustering Horizontal collaborative clustering Knowledge matching Bipartite graph matching KM algorithm
This work is supported by the National Natural Science Foundation of China (No. 11971065, No. 11571001, No. 11701338).
- 1.Liu, Y., Yu, F., Xu, J.: Horizontal collaborative fuzzy clustering based on similarity matrix. In: 2017 29th Chinese Control And Decision Conference (CCDC). IEEE (2017)Google Scholar
- 5.Cleuziou, G., Exbrayat, M., Martin, L., et al.: CoFKM: a centralized method for multiple-view clustering. In: 2009 Ninth IEEE International Conference on Data Mining. IEEE Computer Society (2009)Google Scholar