The embedded neural network accelerator deploying neural network algorithms needs to quantize the values in the neural network model into low-bit integers through quantization operations. However, the ...