Abstract
Due to the fixed array of competing-layer structure, observations distribution features cannot be well reflected by conventional SOM in two-dimensional plane. Aimed at solving this problem, a novel flexible array SOM algorithm (faSOM) is proposed in this paper. This algorithm can adaptively adjust the positions of competing-layer neurons to keep consistent with position of observations. As a result, the neurons in mapping space can maintain the original observation’ features. The faSOM algorithm is successfully applied in pattern recognition of two artificial datasets and red-spotted stonecrop samples. Both theory analysis and experimental results indicate that faSOM is an effective algorithm which can map observation’s inherent feature quickly and accurately. Compared with conventional SOM algorithm, feature mapping effect of faSOM algorithm is much better, because it resolves a typical problem in the conventional SOM that the structure of mapped dataset in competing-layer is distorted.
Similar content being viewed by others
References
Zhiming W, Shiying X (2000) Some perspectives about research and development of modern traditional Chinese medicine. China J Chin Mater Med 25(4):244–245
Cheng YY, Chen MJ, Tong WD (2003) An approach to comparative analysis of chromatographic fingerprints for assuring the quality of botanical drugs. J Chem Inf Comput Sci 43(3):1068–1076
Qixun W, Yan A, Jinghong G, Yongchang L, Jingyuan B, Shuqing H, Jun Y (2006) Study on fingerprint of rhodiola rosea from the Qingzang Plateau by high performance liquid chromatography. J Southwest Univ Nationalities 32(1):72–77
Lavine BK, Davidson CE, Westover DJ (2004) Spectral pattern recognition using self-organizing maps. J Chem Inf Comput Sci 44:1056–1064
Kohonen T (1982) Self-organized formation of topologically correct feature maps. Biol Cyber 43:59–69
Alahakoon D, Halgamuge SK (2000) Dynamic self-organizing maps with controlled growth for knowledge discovery. IEEE Trans Neural Netw 11(3):601–614
Alahakoon LD (2004) Controlling the spread of dynamic self-organizing maps. Neural Comput Appl 13:168–174
Hsu A, Halgamuge SK (2008) Class structure visualization with semi-supervised growing self-organizing maps. Neurocomputing 71:3124–3130
Villmann T, Bauer H-U (1998) Applications of the growing self-organizing map1. Neurocomputing 21:91–100
Yan X, Chen D, Chen Y, Hu S (2001) SOM integrated with CCA for the feature map and classification of complex chemical patterns. Comput Chem 25(6):597–605
Polzlbauer G, Dittenbach M, Rauber A (2006) Advanced visualization of self-organizing maps with vector fields. Neural Netw 19:911–922
Lau KW, Yin H, Hubbard S (2006) Kernel self-organising maps for classification. Neurocomputing 69:2033–2040
Dong H, Zhang Y, Xiaorong P (2008) A new local PCA-SOM algorithm. Neurocomputing 71:3544–3552
Rubio M, Gimnez V (2003) New methods for self-organising map visual analysis. Neural Comput Appl 12(3/4):142–152
Rousseta P, Guinotb C, Mailletc B (2006) Understanding and reducing variability of SOM neighbourhood structure. Neural Netw 19:838–846
John W. Sammon JR (1969) A nonlinear mapping for data structure analysis. IEEE Trans comput 18(5):401–409
Yin H (2002) ViSOM: a novel method for multivariate data projection and structure visualization. IEEE Trans Neural Netw 13(1):237–243
Yin H (2008) On multidimensional scaling and the embedding of self-organising maps. Neural Netw 21:160–169
Dittenbach M, Rauber A, Merkl D (2002) Uncovering hierarchical structure in data using the growing hierarchical self-organizing map. Neurocomputing 48(1):199–216
Lebart L (2006) Assessing self organizing maps via contiguity analysis. Neural Netw 19:847–854
Rynkiewicz J (2006) Self-organizing map algorithm and distortion measure. Neural Netw 19:830–837
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Song, Xf., Wang, Pp. Flexible array self-organizing map neural network and its application in complex pattern recognition. Neural Comput & Applic 20, 9–16 (2011). https://doi.org/10.1007/s00521-010-0421-5
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-010-0421-5