Nonlinear inference capacity of fiber-optical extreme learning machines

IF 6.6 2区 物理与天体物理 Q1 MATERIALS SCIENCE, MULTIDISCIPLINARY
Sobhi Saeed, Mehmet Müftüoğlu, Glitta R. Cheeran, Thomas Bocklitz, Bennet Fischer, Mario Chemnitz
{"title":"Nonlinear inference capacity of fiber-optical extreme learning machines","authors":"Sobhi Saeed, Mehmet Müftüoğlu, Glitta R. Cheeran, Thomas Bocklitz, Bennet Fischer, Mario Chemnitz","doi":"10.1515/nanoph-2025-0045","DOIUrl":null,"url":null,"abstract":"The intrinsic complexity of nonlinear optical phenomena offers a fundamentally new resource to analog brain-inspired computing, with the potential to address the pressing energy requirements of artificial intelligence. We introduce and investigate the concept of nonlinear inference capacity in optical neuromorphic computing in highly nonlinear fiber-based optical Extreme Learning Machines. We demonstrate that this capacity scales with nonlinearity to the point where it surpasses the performance of a deep neural network model with five hidden layers on a scalable nonlinear classification benchmark. By comparing normal and anomalous dispersion fibers under various operating conditions and against digital classifiers, we observe a direct correlation between the system’s nonlinear dynamics and its classification performance. Our findings suggest that image recognition tasks, such as MNIST, are incomplete in showcasing deep computing capabilities in analog hardware. Our approach provides a framework for evaluating and comparing computational capabilities, particularly their ability to emulate deep networks, across different physical and digital platforms, paving the way for a more generalized set of benchmarks for unconventional, physics-inspired computing architectures.","PeriodicalId":19027,"journal":{"name":"Nanophotonics","volume":"25 1","pages":""},"PeriodicalIF":6.6000,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Nanophotonics","FirstCategoryId":"101","ListUrlMain":"https://doi.org/10.1515/nanoph-2025-0045","RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"MATERIALS SCIENCE, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0

Abstract

The intrinsic complexity of nonlinear optical phenomena offers a fundamentally new resource to analog brain-inspired computing, with the potential to address the pressing energy requirements of artificial intelligence. We introduce and investigate the concept of nonlinear inference capacity in optical neuromorphic computing in highly nonlinear fiber-based optical Extreme Learning Machines. We demonstrate that this capacity scales with nonlinearity to the point where it surpasses the performance of a deep neural network model with five hidden layers on a scalable nonlinear classification benchmark. By comparing normal and anomalous dispersion fibers under various operating conditions and against digital classifiers, we observe a direct correlation between the system’s nonlinear dynamics and its classification performance. Our findings suggest that image recognition tasks, such as MNIST, are incomplete in showcasing deep computing capabilities in analog hardware. Our approach provides a framework for evaluating and comparing computational capabilities, particularly their ability to emulate deep networks, across different physical and digital platforms, paving the way for a more generalized set of benchmarks for unconventional, physics-inspired computing architectures.
光纤极限学习机的非线性推理能力
非线性光学现象的内在复杂性为模拟脑启发计算提供了一种全新的资源,具有解决人工智能紧迫的能量需求的潜力。我们在高度非线性的光纤光学极限学习机中引入并研究了非线性推理能力在光学神经形态计算中的概念。我们证明了这种能力随着非线性的扩展,在可扩展的非线性分类基准上,它超过了具有五个隐藏层的深度神经网络模型的性能。通过比较不同工作条件下正常和异常色散光纤以及数字分类器,我们观察到系统的非线性动力学与其分类性能之间存在直接关联。我们的研究结果表明,图像识别任务,如MNIST,在模拟硬件中展示深度计算能力是不完整的。我们的方法提供了一个评估和比较计算能力的框架,特别是它们在不同物理和数字平台上模拟深度网络的能力,为非常规的、物理启发的计算架构的更广泛的基准设置铺平了道路。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
Nanophotonics
Nanophotonics NANOSCIENCE & NANOTECHNOLOGY-MATERIALS SCIENCE, MULTIDISCIPLINARY
CiteScore
13.50
自引率
6.70%
发文量
358
审稿时长
7 weeks
期刊介绍: Nanophotonics, published in collaboration with Sciencewise, is a prestigious journal that showcases recent international research results, notable advancements in the field, and innovative applications. It is regarded as one of the leading publications in the realm of nanophotonics and encompasses a range of article types including research articles, selectively invited reviews, letters, and perspectives. The journal specifically delves into the study of photon interaction with nano-structures, such as carbon nano-tubes, nano metal particles, nano crystals, semiconductor nano dots, photonic crystals, tissue, and DNA. It offers comprehensive coverage of the most up-to-date discoveries, making it an essential resource for physicists, engineers, and material scientists.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信