TY - GEN
T1 - Missing Rule Recovery-based Dynamic Fuzzy Rule Interpolation
AU - Xu, Ruilin
AU - Lin, Jinle
AU - Shang, Changjing
AU - Zhou, Mou
AU - Shen, Qiang
N1 - Publisher Copyright:
© 2026 IEEE.
PY - 2026
Y1 - 2026
N2 - Fuzzy Rule Interpolation (FRI) provides a robust mechanism for deriving conclusions in sparse fuzzy rule bases, particularly when observations do not match existing rules due to insufficient data or incomplete domain expertise. While traditional FRI methods enable approximate inference over the knowledge gaps, they typically discard the interpolated rules immediately after use. This ephemeral nature limits the ability of the underlying fuzzy system to learn and improve efficiency over time. To address this limitation, the concept of Dynamic FRI (D-FRI) has been introduced in the recent literature with an aim to promote interpolated rules into the permanent rule base. Inspired by such initiatives, this paper proposes a novel approach, termed Missing Rule Recovery-based Fuzzy Rule Interpolation (M-FRI). It enhances the performance of sparse rule bases by actively detecting and calculating missing rules during a fuzzy inference process, thereby progressively expanding knowledge coverage in the relevant problem domain. Experimental results demonstrate that the proposed M-FRI method achieves superior performance regarding inference accuracy, as compared to both traditional FRI and existing D-FRI methods.
AB - Fuzzy Rule Interpolation (FRI) provides a robust mechanism for deriving conclusions in sparse fuzzy rule bases, particularly when observations do not match existing rules due to insufficient data or incomplete domain expertise. While traditional FRI methods enable approximate inference over the knowledge gaps, they typically discard the interpolated rules immediately after use. This ephemeral nature limits the ability of the underlying fuzzy system to learn and improve efficiency over time. To address this limitation, the concept of Dynamic FRI (D-FRI) has been introduced in the recent literature with an aim to promote interpolated rules into the permanent rule base. Inspired by such initiatives, this paper proposes a novel approach, termed Missing Rule Recovery-based Fuzzy Rule Interpolation (M-FRI). It enhances the performance of sparse rule bases by actively detecting and calculating missing rules during a fuzzy inference process, thereby progressively expanding knowledge coverage in the relevant problem domain. Experimental results demonstrate that the proposed M-FRI method achieves superior performance regarding inference accuracy, as compared to both traditional FRI and existing D-FRI methods.
KW - dynamic fuzzy reasoning
KW - fuzzy rule interpolation
KW - Missing rule recovery
KW - rule conflict resolution
KW - sparse rule base
UR - https://www.scopus.com/pages/publications/105047275049
U2 - 10.1109/FUZZ69877.2026.11626357
DO - 10.1109/FUZZ69877.2026.11626357
M3 - Conference Proceeding (ISBN)
AN - SCOPUS:105047275049
T3 - 2026 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2026 - Conference Proceedings: Part of the 2026 IEEE World Congress on Computational Intelligence, IEEE WCCI 2026
BT - 2026 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2026 - Conference Proceedings
PB - Institute of Electrical and Electronics Engineers
T2 - 2026 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2026
Y2 - 21 June 2026 through 26 June 2026
ER -