Abstract—Fuzzy cognitive map (FCM) belongs to one of the
soft computing technique for modeling complex systems, which
utilize the advantages from the synergistic theories of neural
networks and fuzzy logic. The development of FCM highly
relies on the human expert experience and knowledge. So,
without those from expert(s), the FCM is hard to be constructed
successfully. In this study, a self-adaptive FCM without any
involvement of experts by using hybrid evolutionary
computation approach is proposed. It includes the genetic
algorithm (GA) and particle swarm optimization (PSO). The
purpose of GA is to decide the significant variables. Based on
those variables selected by GA, the most appropriate cognitive
map can be constructed by PSO, i.e., the relationship matrix for
the set of variables. The purpose of the research is to find the
minimum subset of cognitive variables and the corresponding
correlation matrix from historical numerical dataset so as to
construct the optimal FCM decision model. In this study, the
diagnosis of traditional Chinese medicines has been investigated
base on twelve-meridian data obtained by meridian energy
analysis device. The computational results show that the
proposed approach is able to provide higher classification
accuracy than those of the approaches in literature or by using
commercial software.
Index Terms—Fuzzy cognitive map, genetic algorithm,
particle swarm optimization, traditional Chinese medicine,
twelve-meridian.
T.-C. Chen, C.-H. Wu, and S.-L. Lin are with the Department of
Information Management, National Formosa University, Yunlin, 63201,
Taiwan (e-mail: tchen@ nfu.edu.tw, melody@ nfu.edu.tw,
author@nrim.go.jp).
P.-S. You is now with the Graduate Institute of Marketing and Logistics
Management, National Chiayi University, Chiayi, Taiwan (e-mail:
psyuu@mail.ncyu.edu.tw).
Cite: T.-C. Chen, P.-S. You, C.-H. Wu, and S.-L. Lin, "Using FCM Based Hybrid Computational Approach for Diseases Diagnosis in Traditional Chinese Medicine," International Journal of Machine Learning and Computing vol.4, no. 4, pp. 389-393, 2014.