A discussion on interpretability of linguistic rule based systems and its application to solve regression problems

摘要: Hedge algebras establish a sound approach (HA-approach) to the inherent semantics of words and provides formalized basis incorporate fuzzy set based with their own semantics. On this basis, we are able examine in study new concept semantics-based interpretability FRBSs develop genetic method design interpretable solve regression problems, for instance. The proposed is characterized by following features: (i) guaranteed constraints preserve essential characteristics that can be handled only adjusting fuzziness parameters variables; (ii) Each variable associated word-set rich enough considered as user word-vocabulary variable, called user's Linguistic Frame Cognitive (LFoC), which properly represented multi-granularity structure; (iii) Large cardinalities LFoCs do not increase search space, despite still decrease significantly number initial rules, they generated from patterns given datasets, exploiting similarity-intervals words; (iv) Concurrent learning rule bases determine LFoCs; (v) ability reach suitable trade-offs between word generality specificity accuracy designed FRBSs. shown statistically outperform two counterpart methods examined Alcala et al. (2009), Antonelli (2011) run over 9 6 respectively.