An interval type-2 fuzzy inference system and its meta-cognitive learning algorithm

In this paper, we propose an evolving interval type-2 neuro-fuzzy inference system and its meta-cognitive learning algorithm. The antecedent of each rule is an interval type-2 fuzzy set and the consequent realizes the Takagi–Sugeno–Kang inference mechanism. It employs a data driven mechanism for type-reduction and uncertainty is modeled using both the antecedent and consequent parameters. The learning starts with zero rules and evolves the network parameters and structure as a new sample arrives, using a meta-cognitive learning algorithm. The meta-cognitive learning algorithm regulates learning by employing adequate learning strategies. It uses gradient-descent to adapt the network parameters. The proposed algorithm is referred to as meta-cognitive interval type-2 neuro-fuzzy inference system-gradient descent (McIT2FIS-GD). The performance of McIT2FIS-GD is evaluated using a set of benchmark and a real-world wind speed prediction problem and compared with existing type-1 and type-2 systems. The results highlight that performance of McIT2FIS-GD is superior than other compared algorithms.