IEEE Transactions on Knowledge and Data Engineering
| Hi-index | 0.00 |
Cheung [2] has recently proposed a general learning framework, namely Maximum Weighted Likelihood (MWL), in which an adaptive Rival Penalized EM (RPEM) algorithm has been successfully developed for density mixture clustering with automatic model selection. Nevertheless, its convergence speed relies on the value of learning rate. In general, selection of an appropriate learning rate is a nontrivial task. To circumvent such a selection, this paper further studies the MWL learning framework, and develops a batch RPEM algorithm accordingly provided that all observations are available before the learning process. Compared to the adaptive RPEM algorithm, this batch RPEM need not assign the learning rate analogous to the EM, but still preserve the capability of automatic model selection. Further, the convergence speed of this batch RPEM is faster than the EM and the adaptive RPEM. The experiments show the efficacy of the proposed algorithm.