An In-Memory-Computing Charge-Domain Ternary CNN Classifier Xiangxing Y ang , Member , IEEE

Abstract

The article presents a charge-domain comput- ing ternary neural network (TNN) classifier with a complete four-layer neural network (NN) on a chip. The proposed ternary network provides 1.5-b resolution (0/+1/−1) for weights and acti- vations, leading to 3.9× fewer operations (OPs) per inference than binary neural network (BNN) for the same Modified National Institute of Standards and Technology (MNIST) accuracy. The 1.5-b multiply-and-accumulate (MAC) is implemented by V CM- based capacitor switching scheme, which inherently benefits from the reduced signal swing on the capacitive digital-to-analog converter (CDAC). Also, theV CM-based MAC introduces sparsity during training, resulting in a lower switching rate. The prototype is fabricated in a 40-nm LP CMOS process with an active area of 0.98 mm 2, operates at 549 frames/s (FPS), and consumes 96 µW. With all OPs on the chip, it achieves 97.1% MNIST accuracy with 0.18 µJ per classification, which is the smallest to our knowledge for comparable MNIST classification accuracy.