聚類分析中競爭學習的一種新算法
A NEW COMPETITIVE LEARNING ALGORITHM FOR CLUSTERING ANALYSIS
-
摘要: 分析指出RPCL算法的不足,提出一種競爭學習新算法。新算法引入數(shù)據(jù)點的密度定義,在權(quán)值的調(diào)整中考慮了數(shù)據(jù)集的幾何結(jié)構(gòu)對權(quán)值調(diào)整的影響,克服了RPCL算法的不足。理論分析與實驗表明:新算法不僅可以自動確定數(shù)據(jù)集的類數(shù),而且提高了聚類準確性和收斂速度。
-
關鍵詞:
- 聚類分析; 競爭學習; 密度
Abstract: Based on the analysis of the defect of the RPCL, a new competitive learning algorithm is proposed. In the new algorithm the data density is introduced, and the modification of the weights is taken into account to surmount the defect of the RPCL. It is shown by the theoretical analysis and experimental results that the new algorithm can automatically select the appropriate number of the clusters in a data set, and improve the clustering accuracy and convergence speed. -
Rumelhart D E，Zipser D Feature discovery by competitive learning Cognitive Science，1985.9(1)：75-112．[2]Grossberg S．Competitive learning：from．Iterative activation to adaptive resonance. Cognitive Science，1987：11(1)：23-63．[3]Ahalt S C, Krishnamurty A K，Chen P，Meltion D E.Cmnpetitive learning algorithms for vector quantization[J].Neural Networks.1990,3(2):277-291[4]Xu L，Krzyzak A, Oja E．Riral penalized competitive learning for clustering analysis，RBF net，abd curve detection．IEEE Trans.on Neural Networks、1993：4(4)：636-649． -
計量
- 文章訪問數(shù): 2293
- HTML全文瀏覽量: 131
- PDF下載量: 499
- 被引次數(shù): 0
下載:
