Photonic Integrated Reconfigurable Linear Processors as Neural Network Accelerators

Reconfigurable linear optical processors can be used to perform linear transformations and are instrumental in effectively computing matrix-vector multiplications required in each neural network layer. In this paper, we characterize and compare two thermally tuned photonic integrated processors realized in silicon-on-insulator and silicon nitride platforms suited for extracting feature maps in convolutional neural networks. The reduction in bit resolution when crossing the processor is mainly due to optical losses, in the range 2.3-3.3 for the silicon-on-insulator chip and in the range 1.3-2.4 for the silicon nitride chip. However, the lower extinction ratio of Mach-Zehnder elements in the latter platform limits their expressivity (i.e., the capacity to implement any transformation) to 75%, compared to 97% of the former. Finally, the silicon-on-insulator processor outperforms the silicon nitride one in terms of footprint and energy efficiency.

Automated summary

This paper characterizes and compares two thermally tuned photonic integrated processors, based on silicon-on-insulator and silicon nitride platforms, for feature map extraction in convolutional neural networks. The reduction in bit resolution when crossing the processor is mainly due to optical losses, with the silicon-on-insulator chip ranging from 2.3-3.3 and the silicon nitride chip ranging from 1.3-2.4. However, the lower extinction ratio of Mach-Zehnder elements in the latter platform limits their expressivity to 75% compared to the former. Ultimately, the silicon-on-insulator processor outperforms the silicon nitride one in terms of footprint and energy efficiency.