This paper proposes a novel way of matching general type-2 fuzzy sets using a sequence-based approach. General sequences are defined as an ordered list of objects, which are called events. In our contribution, an event of the sequenced type-2 fuzzy set is defined as the footprint of uncertainty of a specific α-plane. Suited matching algorithms for generalized sequences can be applied to this new interpretation, enabling the computation of the overall dissimilarity value between input type-2 fuzzy sets. The evaluation of the proposed matching method is carried out in the setting of classification, by defining datasets of general type-2 fuzzy sets conceived as labeled patterns. Results show that the methodology is robust, accurate, and computationally acceptable. © 2013 IEEE.
Aggregating α-planes for Type-2 fuzzy set matching / Livi, Lorenzo; Hooman, Tahayori; Alireza, Sadeghian; Rizzi, Antonello. - (2013), pp. 860-865. (Intervento presentato al convegno 9th Joint World Congress on Fuzzy Systems and NAFIPS Annual Meeting, IFSA/NAFIPS 2013 tenutosi a Edmonton; Canada nel 24 June 2013 through 28 June 2013) [10.1109/ifsa-nafips.2013.6608513].
Aggregating α-planes for Type-2 fuzzy set matching
LIVI, LORENZO;RIZZI, Antonello
2013
Abstract
This paper proposes a novel way of matching general type-2 fuzzy sets using a sequence-based approach. General sequences are defined as an ordered list of objects, which are called events. In our contribution, an event of the sequenced type-2 fuzzy set is defined as the footprint of uncertainty of a specific α-plane. Suited matching algorithms for generalized sequences can be applied to this new interpretation, enabling the computation of the overall dissimilarity value between input type-2 fuzzy sets. The evaluation of the proposed matching method is carried out in the setting of classification, by defining datasets of general type-2 fuzzy sets conceived as labeled patterns. Results show that the methodology is robust, accurate, and computationally acceptable. © 2013 IEEE.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.