An efficient morlet function based adaptive method for faster backpropagation for handwritten character recognition
M. Sornam, Muthu Subash Kavitha, M. Poornima Devi · 2016
A Morlet function based adaptive backpropagation learning model was proposed for Tamil handwritten character recognition to increase the convergence speed and minimize the error. In this study the individual adaptive learning rate of both output and hidden layers in the network is adapted in each iteration based on the estimated error values of the network layers. It leads to an efficient and small computation time for learning every change of update values and weights. The promising potential of the Morlet based adaptive backpropagation is shown in comparison to other activation functions and other competing methods.