(a) A single-layer artificial neural network composed of 784 input and 10 output neurons, jointed by independent synapse with different synaptic weight. Each input neuron corresponds to a pixel of the input pattern. (b), (c) Comparisons of the image recognition accuracy at different optoelectronic cases.^[51] Copyright (2019) John Wiley & Sons, Inc. (d) Optic-neural network for recognition of 28 by 28 RGB-colored images. (e) Examples of the training and the testing datasets consisting of single-colored and color-mixed numeric pattern images, respectively.^[63] Copyright (2018) Springer Nature.