New Recurrent Weighted Lyapunov-Stability Based Brain Emotional Learning-Based Neural Network: Application to the Modeling of the Nonlinear Dynamical System

In this paper, a recurrent version of the classical Brain-Emotional Learning Neural Network (BELNN) called as Recurrent Weighted Brain Emotional Learning-Based Neural Network (RWBELNN) is proposed and is applied for the modeling of the nonlinear dynamical system. The proposed model involves an additional tunable connection weights linking the signals of input layers to their corresponding summing junctions present in the Orbitofrontal Cortex (OFC) and Amygdala. Further, tunable recurrent connections are included in both the OFC and Amygdala components that form the link between the unit-delayed outputs of these respective components to their corresponding summing junctions. These weighted connections improves the memory component of the model which in turn improves its nonlinear dynamic system handling. The proposed model also offers a relatively lower computational complexity as compared to other recurrent neural models because of its simpler structure as compared to the other recurrent neural network models available in the literature. To develop the parameter update equations and to ensure the model’s overall stability, Lyapunov-stability principles are invoked. The performance of the proposed model is compared with state-of-the art neural models and is found to be superior.