A Study on New Performance Index of Granular Models

다국어 초록 (Multilingual Abstract)

In this paper, we propose the performance evaluation of granular model instead of the conventional performance index, which is a root mean square error. In contrast to most fuzzy models developed in the previous researches, the outputs obtained by granular models have the form of information granules rather than plain numeric entities. Thus, we use a concept of information granularity as a network of information granules with contexts produced in the output space and a set of information granules formed in the input space. Granular model is constructed by information granules that are extracted by specialized context-based fuzzy clustering. Furthermore, this model is designed by the rules connecting the association of information granules obtained in input-output space. The performance evaluation method of granular model needs new criterion, because the output of granular model has triangular fuzzy number. The performance index is expressed by the product between fuzzy coverage of predicted output and specificity of information granules. Thus, we can quantify the model output representing information granules. Granular models offer a new performance evaluation method of system modeling by designing models at the level of information granules. For this, we deal with a synthesis function approximation and nonlinear regression and demonstrate the effectiveness of the proposed performance evaluation in the design of granular model.

참고문헌 (Reference)

1 이명원, "인간 중심형 시스템 및 컴퓨팅을 위한 점진적인 방사기저함수 네트워크의 설계" 한국정보기술학회 10 (10): 137-142, 2012

2 W. Pedrycz, "The design of free structure granular mapping : The use of the principle of justifiable granularity" 43 (43): 2105-2113, 2013

3 C. U. Yeom, "Performance index method of granular model based on information granules" 220-221, 2016

4 W. Pedrycz, "Linguistic models and linguistic modeling" 29 (29): 601-612, 1998

5 W. Pedrycz, "Handbook of Granular Computing" John Wiley & Sons 2008

6 O. F. Reyes-Galaviz, "Granular fuzzy models : analysis, design, and evaluation" 64 : 1-19, 2015

7 S. S. Kim, "Development of quantum-based adaptive neuro-fuzzy networks" 40 (40): 91-100, 2010

8 W. Pedrycz, "Designing fuzzy sets with the use of the parametric principle of justifiable granularity" 24 (24): 489-496, 2016

9 W. Pedrycz, "Conditional fuzzy c-means" 17 (17): 625-631, 1996

10 곽근창, "A Design of Granular Model Using Conditional Clustering and Density Peaks" 한국정보기술학회 13 (13): 127-134, 2015