2015 IEEE International Conference on Data Mining (ICDM) (2015)
Atlantic City, NJ, USA
Nov. 14, 2015 to Nov. 17, 2015
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/ICDM.2015.48
The parameter estimation to mixture models has been shown as a local optimal solution for decades. In this paper, we propose a functional estimation to mixture models using step functions. We show that the proposed functional inference yields a convex formulation and consequently the mixture models are feasible for a global optimum inference. The proposed approach further unifies the existing isolated exemplar-based clustering techniques at a higher level of generality, e.g. it provides a theoretical justification for the heuristics of the clustering by affinity propagation Frey & Dueck (2007), it reproduces Lashkari & Golland (2007)'s's convex formulation as a special case under this step function construction. Empirical studies also verify the theoretic justifications.
Mixture models, Function approximation, Estimation, Bayes methods, Electronic mail, Inference algorithms
Y. Xu, Y. Zhu, Z. Zhang, Y. Zhang and P. S. Yu, "Convex Approximation to the Integral Mixture Models Using Step Functions," 2015 IEEE International Conference on Data Mining (ICDM), Atlantic City, NJ, USA, 2015, pp. 479-488.