The Mahalanobis Distance Based Rival Penalized Competitive Learning Algorithm
نویسندگان
چکیده
The rival penalized competitive learning (RPCL) algorithm has been developed to make the clustering analysis on a set of sample data in which the number of clusters is unknown, and recent theoretical analysis shows that it can be constructed by minimizing a special kind of cost function on the sample data. In this paper, we use the Mahalanobis distance instead of the Euclidean distance in the cost function computation and propose the Mahalanobis distance based rival penalized competitive learning (MDRPCL) algorithm. It is demonstrated by the experiments that the MDRPCL algorithm can be successful to determine the number of elliptical clusters in a data set and lead to a good classification result.
منابع مشابه
A New Competitive Learning Algorithm for Data Clustering
This paper presents a new competitive learning algorithm for data clustering, named the dynamically penalized rival competitive learning algorithm (DPRCA). It is a variant of the rival penalized competitive algorithm [1] and it performs appropriate clustering without knowing the clusters number, by automatically driving extra seed points far away from the input data set. It doesn’t have the "de...
متن کاملExpectation-MiniMax: A General Penalized Competitive Learning Approach to Clustering Analysis
In the literature, the Rival Penalized Competitive Learning (RPCL) algorithm (Xu et al. 1993) and its variants perform clustering analysis well without knowing the cluster number. However, such a penalization scheme is heuristically proposed without any theoretical guidance. In this paper, we propose a general penalized competitive learning approach named Expectation-MiniMax (EMM) Learning that...
متن کاملConvergence Analysis of Rival Penalized Competitive Learning (RPCL) Algorithm
This paper analyzes the convergence of the Rival Penalized Competitive Learning (RPCL) algorithm via a cost function. It is shown that as RPCL process decreases the cost to a global minimum, a correct number of weight vectors will converge to each center of the clusters in the sample data, respectively, while the others diverge.
متن کاملCompetitive Learning Algorithms for Data Clustering
This paper presents and discusses some competitive learning algorithms for data clustering. A new competitive learning algorithm, named the dynamically penalized rival competitive learning algorithm (DPRCL), is introduced and studied. It is a variant of the rival penalized competitive algorithm [1] and it performs appropriate clustering without knowing the clusters number, by automatically driv...
متن کاملRival Penalized Competitive Learning Based Separator on Binary Sources Separation
This paper 1 presents an approach named Rival Penalized Competitive Learning based Binary Source Separator (RPCL-BSS), which has two major advantages: (1) fast in implementation , (2) able to automatically determine the number of binary sources, and (3) able to reduce the noise eeects. Experiments have shown that RPCL-BSS algorithm can not only nd out the correct number of sources quickly, but ...
متن کامل