All‐Optical Nonlinear Activation Functions Based on Photonic Crystal Microcavity
- 15 September 2026
- journal article
- Published by Wiley in Laser & Photonics Reviews
Abstract
The growing computational demand of artificial intelligence (AI) calls for energy‐efficient computing architectures. Diffractive deep neural networks () offer a promising route for optical computing, but their lack of nonlinear activation functions limits their representation capability. Here, we demonstrate a passive all‐optical Leak‐ReLU nonlinear activation function based on a 1D photonic crystal microcavity. By exploiting resonantly enhanced thermo‐optic nonlinearity, the cavity exhibits a temperature‐induced resonance blueshift that produces intensity‐dependent nonlinear transmission with tunable activation thresholds through wavelength selection. Integrating the experimentally characterized nonlinear activation function into a improves the classification accuracy by 1.79% on Fashion‐MNIST compared with a two‐layer linear and accelerates convergence by 2.6 times on MNIST. This work provides a compact and passive approach for implementing nonlinear optical neural networks and advances low‐power neuromorphic optical computing.Keywords
Funding Information
- National Key Research and Development Program of China (2023YFA1407200)
- National Natural Science Foundation of China (12222408)
- National Natural Science Foundation of China (12304423)
- National Natural Science Foundation of China (12304424)
- National Natural Science Foundation of China (12474290)
- National Natural Science Foundation of China (12474385)
- National Natural Science Foundation of China (12547140)
- National Natural Science Foundation of China (12674408)
- National Natural Science Foundation of China (12604503)
- Natural Science Foundation of Tianjin Municipality (24JCJQJC00070)
- Natural Science Foundation of Tianjin Municipality (24JCYBJC01600)
- Natural Science Foundation of Tianjin Municipality (23JCYBJC01210)
- China Postdoctoral Science Foundation (2022M711710)
- China Postdoctoral Science Foundation (2026M793701)
- Fundamental Research Funds for the Central Universities (63261128)
- Fundamental Research Funds for the Central Universities (63263116)
- Fundamental Research Funds for the Central Universities (92612027)
- Natural Science Foundation of Guangxi Province (2024GXNSFAA010314)
- Higher Education Discipline Innovation Project (B23045)
This publication has 33 references indexed in Scilit:
- High-Resolution and Ultralow-Power Nonlinear Image Processing with Passive High-Quality Factor MetasurfacesNano Letters, 2026
- Inverse design and performance prediction of autofocusing beams using deep learningOptics & Laser Technology, 2025
- Optical trapping force exploration on Mie particles for structured light via multivariate sensitivity analysis and deep learningOptics Communications, 2025
- Tunable Nonlinear Activation Functions Enabled by WO₃ Films for Adaptive Diffractive Deep Neural NetworksLaser & Photonics Reviews, 2025
- Highly Intelligent Forward Design of Metamaterials Empowered by Circuit‐Physics‐Driven Deep LearningLaser & Photonics Reviews, 2024
- Efficient Thermo-Optic Switching and Modulating in Nonlocal MetasurfacesNano Letters, 2024
- A Lightweight Neural Network for Spectroscopic Ellipsometry AnalysisAdvanced Optical Materials, 2023
- Bright Second Harmonic Emission from Photonic Crystal Vertical CavityAdvanced Functional Materials, 2023
- Arbitrary Multifunctional Vortex Beam Designed by Deep Neural NetworkAdvanced Optical Materials, 2023
- Ultracompact topological photonic switch based on valley-vortex-enhanced high-efficiency phase shiftLight: Science & Applications, 2022