Meng Xiansong, Kong Deming, Kim Kwangwoong, Li Qiuchi, Dong Po, Cox Ingemar J, Lioma Christina, Hu Hao
DTU Electro, Technical University of Denmark, Kgs. Lyngby, 2800, Denmark.
Nokia Bell Labs, New Providence, NJ, 07974, USA.
Nat Commun. 2025 Aug 12;16(1):7465. doi: 10.1038/s41467-025-62586-0.
Optical neural networks (ONNs) promise computing efficiency beyond microelectronics for modern artificial intelligence (AI). Current ONNs using analog matrix-vector multiplication (MVM) implementations are fundamentally limited in numerical precision due to accumulated noise in electro-optical processing. We propose a digital-analog hybrid MVM architecture that achieves a high numerical precision without sacrificing computing efficiency. Our fabricated proof-of-concept hybrid optical processor (HOP) achieves 16-bit precision in high-definition image processing, with a pixel error rate of 1.8 × 10 at a signal-to-noise ratio of 18.2 dB, and shows no accuracy loss in MNIST digit recognition. We further explore applying the HOP processor in You Look Only Once (YOLO) object detection and demonstrate sufficient numerical precision is crucial for high confidence detection in real-world neural networks. The hybrid optical computing concept may be applied to various photonic MVM implementations to enable accurate optical computing architectures.
光学神经网络（ONNs）有望为现代人工智能（AI）带来超越微电子技术的计算效率。当前使用模拟矩阵向量乘法（MVM）实现的ONNs由于电光处理中累积的噪声，在数值精度上存在根本限制。我们提出了一种数模混合MVM架构，该架构在不牺牲计算效率的情况下实现了高数值精度。我们制造的概念验证混合光学处理器（HOP）在高清图像处理中实现了16位精度，在信噪比为18.2 dB时像素错误率为1.8×10，并且在MNIST数字识别中没有精度损失。我们进一步探索将HOP处理器应用于You Look Only Once（YOLO）目标检测，并证明足够的数值精度对于现实世界神经网络中的高置信度检测至关重要。混合光学计算概念可应用于各种光子MVM实现，以实现精确的光学计算架构。
