Input Voltage Scheme for DOT Product Engine Using NAND Flash Cells

Hardware-based neural networks are expected to be a new computing breakthrough beyond conventional von Neumann architecture because of their low power operations. In this work, we investigate input voltage scheme for feedforward neural networks using NAND flash cells as synaptic devices. Additional periphery circuit is required to apply input to the word line. However, due to the linearity of I
BL
-V
BL
curve in the linear region, input can be applied to bit line without additional circuits.