A Meta-cognitive Recurrent Fuzzy Inference System with Memory Neurons (McRFIS-MN) and its Fast Learning Algorithm for Time Series Forecasting

In this paper, a Meta-cognitive Recurrent Fuzzy Inference System is proposed where recurrence is brought using Memory type Neurons (McRFIS-MN) to retain the effect of all past instances, while the meta-cognition component is employed to control the learning process, by deciding what-to-learn, when-to-learn and how-to-learn from the training data. The McRFIS-MN model has five layers, and Memory Neurons (MN) are employed only in the layers handling crisp values. The antecedent parameters are set randomly while only the consequent weights of the network are updated using a one-shot type projection based learning algorithm through time (PBLT) which makes the learning very fast. The performance evaluation of McRFIS-MN has been carried out using benchmark problems in the areas of nonlinear system identification and time-series forecasting. The results are evaluated against some of the most popular neural fuzzy methods and the obtained results indicate that McRFIS-MN performs better in terms of speed while achieving better or similar accuracy.