ROSA: Robust and Energy-Efficient Microring-Based Optical Neural Networks via Optical Shift-and-Add and Layer-Wise Hybrid Mapping

TL;DR

ROSA is a microring-based optical neural network architecture that enhances robustness and energy efficiency through innovative modules and strategies, achieving significant reductions in energy-delay product and improved accuracy.

Contribution

It introduces a noise-aware voltage-to-weight model, an optical shift-and-add module, and a layer-wise hybrid mapping strategy for robust, energy-efficient optical neural networks.

Findings

Reduced energy-delay product by up to 64% compared to baseline architectures.

Improved CIFAR-10 accuracy by 8.3% over weight-stationary mapping.

Achieved 54.7% lower EDP than DEAP-CNNs on average.

Abstract

This work presents ROSA, a microring-based optical neural network architecture that improves robustness and energy efficiency using an optical shift-and-add (OSA) module and a layer-wise hybrid mapping strategy. It introduces a noise-aware voltage-to-weight model considering DAC and thermal variations, and a workload-aware framework to co-optimize MRR array size and layer-wise dataflow. Optimized arrays reduce the aggregated relative energy-delay product (EDP) by 64% and 26% compared with DEAP-CNNs and a general compact array, respectively. OSA further contributes 29% EDP reduction. The proposed hybrid mapping strategy improves CIFAR-10 accuracy by 8.3% over weight-stationary mapping while achieving an average 54.7% lower EDP than DEAP-CNNs.