Loading…
Learning rules for Sugeno ANFIS with parametric conjunction operations
The paper presents a Sugeno Adaptive Neuro-Fuzzy Inference System with parametric conjunction operations architecture, ANFIS-CX. The advantages of using parametric conjunction operations in fuzzy models are discussed, and learning rules for system identification with such operations are proposed. Th...
Saved in:
Published in: | Applied soft computing 2020-04, Vol.89, p.106095, Article 106095 |
---|---|
Main Authors: | , , , , , , , |
Format: | Article |
Language: | English |
Subjects: | |
Citations: | Items that this one cites Items that cite this one |
Online Access: | Get full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | The paper presents a Sugeno Adaptive Neuro-Fuzzy Inference System with parametric conjunction operations architecture, ANFIS-CX. The advantages of using parametric conjunction operations in fuzzy models are discussed, and learning rules for system identification with such operations are proposed. These learning strategies can include steepest descent gradient, differential evolution and least square estimation algorithms for tuning antecedent, conjunction, and consequent parameters, respectively. The results of system identification by parameter tuning of conjunction operations in addition to or instead of parameter tuning of the input membership functions are presented. Simulation results show that parameter training in conjunction operations, composed of four basic t-norms, significantly improves the approximation capability of fuzzy models.
•Sugeno ANFIS architecture with parametric conjunction operations is introduced.•A learning rule is proposed using hybrid optimization techniques: SD, LSE, and DEA.•Basic t-norms operations are used to generate fuzzy parametric conjunctions.•We study the domain regions of fuzzy systems sensitive to the conjunction operations.•It was proved parametric conjunction operations improve ANFIS performance. |
---|---|
ISSN: | 1568-4946 1872-9681 |
DOI: | 10.1016/j.asoc.2020.106095 |