A design of neural network controller based on autotuning the gain of the activation function of neurons is accomplished. Such a gain-tuning procedure is combined with the conventional weight-tuning backpropagation algorithm in the learning phase to provide more efficient and faster learning of a neural network. Satisfactory results are obtained when using this method to control a nonlinear plant.
'/%3e%3cuse xlink:href='%23a' transform='rotate(60)'/%3e%3ccircle r='17' fill='%23000095'/%3e%3ccircle r='15' fill='%23fff'/%3e%3c/svg%3e)