George Sarantoglou;Adonis Bogris;Charis Mesaritakis
{"title":"通过相位调制实现全光学、可重构和功率无关的神经激活功能","authors":"George Sarantoglou;Adonis Bogris;Charis Mesaritakis","doi":"10.1109/JQE.2024.3437353","DOIUrl":null,"url":null,"abstract":"In this work, we present numerical results concerning an integrated photonic non-linear activation function that relies on a power independent, non-linear phase to amplitude conversion in a passive optical resonator. The underlying mechanism is universal to all optical filters, whereas here, simulations were based on micro-ring resonators. Investigation revealed that the photonic neural node can be tuned to support a wide variety of continuous activation functions that are relevant to the neural network architectures, such as the sigmoid and the soft-plus functions. The proposed photonic node is numerically evaluated in the context of time delayed reservoir computing (TDRC) scheme, targeting the one-step ahead prediction of the Santa Fe series. The proposed phase to amplitude TDRC is benchmarked versus the conventional amplitude based TDRC, showcasing a performance boost by one order of magnitude.","PeriodicalId":13200,"journal":{"name":"IEEE Journal of Quantum Electronics","volume":null,"pages":null},"PeriodicalIF":2.2000,"publicationDate":"2024-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"All-Optical, Reconfigurable, and Power Independent Neural Activation Function by Means of Phase Modulation\",\"authors\":\"George Sarantoglou;Adonis Bogris;Charis Mesaritakis\",\"doi\":\"10.1109/JQE.2024.3437353\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"In this work, we present numerical results concerning an integrated photonic non-linear activation function that relies on a power independent, non-linear phase to amplitude conversion in a passive optical resonator. The underlying mechanism is universal to all optical filters, whereas here, simulations were based on micro-ring resonators. Investigation revealed that the photonic neural node can be tuned to support a wide variety of continuous activation functions that are relevant to the neural network architectures, such as the sigmoid and the soft-plus functions. The proposed photonic node is numerically evaluated in the context of time delayed reservoir computing (TDRC) scheme, targeting the one-step ahead prediction of the Santa Fe series. The proposed phase to amplitude TDRC is benchmarked versus the conventional amplitude based TDRC, showcasing a performance boost by one order of magnitude.\",\"PeriodicalId\":13200,\"journal\":{\"name\":\"IEEE Journal of Quantum Electronics\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":2.2000,\"publicationDate\":\"2024-08-02\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"IEEE Journal of Quantum Electronics\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://ieeexplore.ieee.org/document/10620692/\",\"RegionNum\":3,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"ENGINEERING, ELECTRICAL & ELECTRONIC\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE Journal of Quantum Electronics","FirstCategoryId":"5","ListUrlMain":"https://ieeexplore.ieee.org/document/10620692/","RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"ENGINEERING, ELECTRICAL & ELECTRONIC","Score":null,"Total":0}
引用次数: 0
摘要
在这项工作中,我们展示了有关集成光子非线性激活函数的数值结果,该函数依赖于无源光学谐振器中与功率无关的非线性相位到振幅转换。其基本机制适用于所有光学滤波器,而这里的模拟则基于微环谐振器。研究表明,光子神经节点可以进行调整,以支持与神经网络架构相关的各种连续激活函数,如sigmoid函数和软加函数。在时延储层计算(TDRC)方案的背景下,对所提出的光子节点进行了数值评估,目标是提前一步预测 Santa Fe 序列。拟议的相位到振幅 TDRC 与传统的基于振幅的 TDRC 相比,性能提高了一个数量级。
All-Optical, Reconfigurable, and Power Independent Neural Activation Function by Means of Phase Modulation
In this work, we present numerical results concerning an integrated photonic non-linear activation function that relies on a power independent, non-linear phase to amplitude conversion in a passive optical resonator. The underlying mechanism is universal to all optical filters, whereas here, simulations were based on micro-ring resonators. Investigation revealed that the photonic neural node can be tuned to support a wide variety of continuous activation functions that are relevant to the neural network architectures, such as the sigmoid and the soft-plus functions. The proposed photonic node is numerically evaluated in the context of time delayed reservoir computing (TDRC) scheme, targeting the one-step ahead prediction of the Santa Fe series. The proposed phase to amplitude TDRC is benchmarked versus the conventional amplitude based TDRC, showcasing a performance boost by one order of magnitude.
期刊介绍:
The IEEE Journal of Quantum Electronics is dedicated to the publication of manuscripts reporting novel experimental or theoretical results in the broad field of the science and technology of quantum electronics. The Journal comprises original contributions, both regular papers and letters, describing significant advances in the understanding of quantum electronics phenomena or the demonstration of new devices, systems, or applications. Manuscripts reporting new developments in systems and applications must emphasize quantum electronics principles or devices. The scope of JQE encompasses the generation, propagation, detection, and application of coherent electromagnetic radiation having wavelengths below one millimeter (i.e., in the submillimeter, infrared, visible, ultraviolet, etc., regions). Whether the focus of a manuscript is a quantum-electronic device or phenomenon, the critical factor in the editorial review of a manuscript is the potential impact of the results presented on continuing research in the field or on advancing the technological base of quantum electronics.