Figure S21. Training of “K”, “N”, “U”, “A”, and “I” letters to the developed neural networks. (a) A flow chart for training logics to the developed neural network. (b) The ideal weight distribution of the synapse cells for the operation of the letters, and the consumed energy for training each letter. For training “K”, “N”, “U”, “A”, and “I” letters, about 1.68, 2.45, 2.40, 0.82, and 1.14 nJ were consumed, respectively.