arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2607.08360physics.opticseess.SP
收藏

用于多任务近场光子计算的逆设计元处理单元

Inverse-designed meta processing units for multi-task near-field photonic computing

Tsinghua University(清华大学);Zhangjiang Laboratory(张江实验室)

Chu Wu, Zeyu Cai, Songtao Yang, Ruoyu Shen, Yinan Zhao, Haiou Zhang, Wei Chu, Xing Lin

AI总结:

研究针对集成光子神经网络需求,提出逆设计的元处理单元(MPU),它能实现局部复矩阵变换。通过实验验证其在矩阵运算等方面性能,并与有源组件及训练集成,在多任务识别中表现出色,确立其为紧凑无源矩阵算子。

摘要:

集成光子神经网络需要同时具备紧凑、矩阵通用且与任务级可重构性兼容的光学算子。本文介绍了一种元处理单元(MPU),它是一种逆设计的近场光子器件,可在浅蚀刻硅区域内实现局部复矩阵变换。每个2x2算子占据9.6 umx4.8 um,被设计为可重复使用的无源矩阵基元,可与可重构MZI神经元结合。展示了一个3位量化的等效MZI酉器件库,有效重构精度为3.32位。验证了任意复2x2矩阵拟合和级联4x4矩阵运算的保真度为92.7%。还将MPU与有源光子组件和硬件在环训练集成,在双任务元音识别上实现了83.5%和80.9%的测试准确率。在大规模EMNIST模拟中,细粒度神经元级MPU替换策略在90%共享MPU替换时平均准确率达到87.64%,比层级基线高7.26个百分点。这些结果确立了逆设计的MPU作为异构、硬件自适应光子神经网络的紧凑无源矩阵算子。

英文摘要:

Integrated photonic neural networks require optical operators that are simultaneously compact, matrix-general and compatible with task-level reconfigurability. Here we introduce a meta processing unit (MPU), an inverse-designed near-field photonic device that implements local complex matrix transformations within a shallow-etched silicon region. Each 2x2 operator occupies 9.6 umx4.8 um and is designed as a reusable passive matrix primitive that can be combined with reconfigurable MZI neurons. We demonstrate a 3-bit quantized MZI-equivalent unitary device library with an effective reconstruction precision of 3.32 bits. Beyond unitary operators, we validate arbitrary complex 2x2 matrix fitting and a cascaded 4x4 matrix operation with 92.7% fidelity. We further integrate the MPU with active photonic components and hardware-in-the-loop training, achieving test accuracies of 83.5% and 80.9% on dual-task vowel recognition. In large-scale EMNIST simulations, a fine-grained neuron-level MPU replacement strategy reaches 87.64% average accuracy at 90% shared-MPU replacement, outperforming a layer-level baseline by 7.26 percentage points. These results establish inverse-designed MPUs as compact passive matrix operators for heterogeneous, hardware-adaptive photonic neural networks.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑