实现(概率)
荧光
计算机科学
人工神经网络
计算机硬件
嵌入式系统
人工智能
物理
数学
光学
统计
作者
Songrui Wei,Shangcheng Yang,Dingchen Wang,Xiao Tang,Kunbin Huang,Yanqi Ge,Bowen Du,Zhi Chen,Zhongrui Wang,Xiaojun Liang,Weihua Gui,Wen Gao,Dianyuan Fan,Han Zhang
出处
期刊:Optica
[Optica Publishing Group]
日期:2025-05-21
卷期号:12 (7): 968-968
被引量:3
标识
DOI:10.1364/optica.555491
摘要
Artificial neural networks (ANNs) are biomimetic computational frameworks inspired by biological neural systems, consisting of interconnected artificial neurons and synaptic connections. Optical neural networks (ONNs), in turn, are hardware implementations where photonic materials and devices are structured in neuromorphic configurations, utilizing optical signals as the information carrier for computation. They leverage direct optical signal processing, thereby offering unparalleled advantages over conventional electronic computing, including low latency, reduced energy consumption, and exceptional parallelism. To date, existing ONNs are predominantly reliant on physical processes such as interference and absorption. This study proposes fluorescence matrix–vector multiplication (FMVM), a method to realize the linear part of ONNs. It is based on the use of a spiropyran film exhibiting reversible fluorescence modification, which employs photochromism for nonvolatile reconfigurable modulation of fluorescence efficiency as a multiplier. The fluorescence implements the multiplication operation of the input excitation signal and the fluorescent efficiency via photon absorption and emission, which facilitates wavelength transition and ultraviolet (UV) light in-memory-display computing. The system achieves dual functionality: processing information content while simultaneously transducing its physical carrier from invisible to visible wavelengths. We experimentally validate the programmability and nonvolatility of a fluorescent efficiency-based multiplier via dual-light irradiation. Further, the application of FMVM in UV processing is demonstrated based on a fingerprint identification example, achieving passive and nano-second latency processing and display capability. In this work, fluorescence is suggested as a way to realize ONN. It has special advantages in displaying, energy consuming, and latency in treating optical signals. It will advance smart sensing, autonomous systems, and medical imaging by delivering compact, user-friendly solutions with unprecedented time and energy efficiency.
科研通智能强力驱动
Strongly Powered by AbleSci AI