A Meta-cognitive Fully Complex valued Fast Learning Predictor network for solving real valued prediction problems

The paper presents a Meta-cognitive Fully Complex-valued Fast Learning Predictor (Mc-FCFLP) for solving real-valued prediction problems. Mc-FCFLP has two components namely, a cognitive component and a meta-cognitive one. The meta-cognitive component of Mc-FCFLP consist of a self-regulatory learning mechanism which fixes what-to-learn, when-to-learn and how-to-learn. As the training samples are provided to the network one by one, the meta-cognitive component chooses appropriate learning strategies for the sample. The training sample is either gets deleted, utilised to add a new neuron or it is hold back for the future use. Hence, the architecture of Mc-FCFLP is build throughout the training process. The performance of the proposed predictor is evaluated compared to the other complex-valued and a some best performing real-valued networks with a set of benchmark Prediction problems. Performance outcomes show that the Mc-FCFLP has better prediction ability compared to the other networks shown in the literature.