Device‐System End‐to‐End Design of Photonic Neuromorphic Processor Using Reinforcement Learning
Device‐System End‐to‐End Design of Photonic Neuromorphic Processor Using Reinforcement Learning
复制标题
使用强化学习的光子神经形态处理器的设备-系统端-端设计
DOI：
10.1002/lpor.202200381
复制
发表时间：
2022
影响因子：
11
通讯作者：
Gao, Weilu
中科院分区：
文献类型：
--
作者：
Tang, Yingheng;Zamani, Princess Tara;Chen, Ruiyang;Ma, Jianzhu;Qi, Minghao;Yu, Cunxi;Gao, Weilu
将高性能光电器件纳入光子神经形态处理器可以大大加速机器学习（ML）算法中的计算密集型矩阵乘法运算。然而，传统的单个设备和系统的设计在很大程度上是断开的，并且系统优化仅限于小设计空间的手动探索。在这里，一个设备系统端到端的设计方法，报告优化自由空间光学通用矩阵乘法（GEMM）硬件加速器的工程由硫系相变材料制成的空间可重构阵列。利用具有实验信息的高度并行化集成硬件仿真器，通过强化学习算法（包括深度Q学习神经网络、贝叶斯优化及其级联方法）探索大的参数空间，实现直接优化GEMM计算精度的单元器件设计。算法生成的物理量显示了系统性能指标与器械规格之间的明确相关性。此外，采用物理感知训练方法来部署优化的硬件，以执行图像分类，材料发现和光学ML加速器的闭环设计任务。演示的框架提供了对光电设备和系统的端到端和协同设计的见解，减少了人为监督和领域知识障碍。
The incorporation of high‐performance optoelectronic devices into photonic neuromorphic processors can substantially accelerate computationally intensive matrix multiplication operations in machine learning (ML) algorithms. However, the conventional designs of individual devices and system are largely disconnected, and the system optimization is limited to the manual exploration of a small design space. Here, a device‐system end‐to‐end design methodology is reported to optimize a free‐space optical general matrix multiplication (GEMM) hardware accelerator by engineering a spatially reconfigurable array made from chalcogenide phase change materials. With a highly parallelized integrated hardware emulator with experimental information, the design of unit device to directly optimize GEMM calculation accuracy is achieved by exploring a large parameter space through reinforcement learning algorithms, including deep Q‐learning neural network, Bayesian optimization, and their cascaded approach. The algorithm‐generated physical quantities show a clear correlation between system performance metrics and device specifications. Furthermore, physics‐aware training approaches are employed to deploy optimized hardware to the tasks of image classification, materials discovery, and a closed‐loop design of optical ML accelerators. The demonstrated framework offers insights into the end‐to‐end and co‐design of optoelectronic devices and systems with reduced human supervision and domain knowledge barriers.
登录
查看更多内容
影响因子：
4.3
作者：
Nandakumar, S. R.;Le Gallo, Manuel;Eleftheriou, Evangelos
通讯作者：
Eleftheriou, Evangelos
影响因子：
16.6
作者：
Wang T;Ma SY;Wright LG;Onodera T;Richard BC;McMahon PL
通讯作者：
McMahon PL
物理 - 复杂的感知机器学习和对抗性攻击 - 有价值的可重构衍射全 - 光神经网络
影响因子：
11
作者：
Chen, Ruiyang;Li, Yingjie;Lou, Minhan;Fan, Jichao;Tang, Yingheng;Sensale‐Rodriguez, Berardi;Yu, Cunxi;Gao, Weilu
通讯作者：
Gao, Weilu
影响因子：
35
作者：
Shen, Yichen;Harris, Nicholas C.;Soljacic, Marin
通讯作者：
Soljacic, Marin
影响因子：
64.8
作者：
Wright LG;Onodera T;Stein MM;Wang T;Schachter DT;Hu Z;McMahon PL
通讯作者：
McMahon PL