Acceleration of Deep Neural Network Training Using Field Programmable Gate Arrays

Figure 8

Energy efficiency with difference data quantization on FPGA.