Photonic Convolution Neural Network Based on Interleaved Time-Wavelength Modulation

Convolution neural network (CNN), as one of the most powerful and popular technologies, has achieved remarkable progress for image and video classification since its invention. However, with the high definition video-data explosion, convolution layers in the CNN architecture will occupy a great amount of computing time and memory resources due to high computation complexity ofmatrix multiply accumulate operation. In this paper, a novel integrated photonic CNN is proposed based on double correlation operations through interleaved time-wavelength modulation. Micro-ring based multi-wavelengthmanipulation and single dispersionmedium are utilized to realize convolution operation and replace the conventional optical delay lines. 200 images are tested in MNIST datasets with accuracy of 85.5% in our photonic CNN versus 86.5% in 64-bit computer. We also analyze the computing error of photonic CNN caused by various micro-ring parameters, operation baud rates and the characteristics of micro-ring weighting bank. Furthermore, a tensor processing unit based on 4x4mesh with 1.2 Tera operation per second (TOPS) computing capability at 20 G baud rate is proposed and analyzed to form a paralleled photonic CNN.

Automated summary

This paper introduces a novel integrated photonic CNN based on double correlation operations and time-wavelength modulation. Testing on the MNIST dataset shows an accuracy of 85.5% for the photonic CNN, slightly lower than 86.5% achieved by a 64-bit computer. The study also analyzes the computing error of the photonic CNN and proposes a parallel photonic CNN based on a tensor processing unit.