All‐Optical Autoencoder Machine Learning Framework Using Linear Diffractive Processors

Diffractive deep neural network (D2NN), known for its high speed and strong parallelism, is applied across various fields, including pattern recognition, image processing, and image transmission. However, existing network architectures primarily focus on data representation within the original domai...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Laser & photonics reviews Ročník 19; číslo 15
Hlavní autori: Feng, Peijie, Tan, Yong, Chong, Mingzhe, Li, Lintao, Zhang, Zongkun, Liu, Fubei, Wen, Yongzheng, Tan, Yunhua
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Weinheim Wiley Subscription Services, Inc 01.08.2025
Predmet:
Artificial neural networks
autoencoder
Decoders
diffractive neural network
Image classification
image denoising
image generation
Image processing
Image reconstruction
Image transmission
Machine learning
Multiplexing
Pattern recognition
Processors
Representations
ISSN:1863-8880, 1863-8899
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:Diffractive deep neural network (D2NN), known for its high speed and strong parallelism, is applied across various fields, including pattern recognition, image processing, and image transmission. However, existing network architectures primarily focus on data representation within the original domain, with limited exploration of the latent space, thereby restricting the information mining capabilities and multifunctional integration of D2NNs. Here, an all‐optical autoencoder (OAE) framework is proposed that linearly encodes the input wavefield into a prior shape distribution in the diffractive latent space (DLS) and decodes the encoded pattern back to the original wavefield. By leveraging the bidirectional multiplexing property of D2NN, the OAE modelsfunction as encoders in one direction and as decoders in the opposite direction. The models are applied to three areas: image denoising, noise‐resistant reconfigurable image classification, and image generation. Proof‐of‐concept experiments are conducted to validate numerical simulations. The OAE framework exploits the potential of latent representations, enabling single set of diffractive processors to simultaneously achieve image reconstruction, representation, and generation. This work not only offers fresh insights into the design of optical generative models but also paves the way for developing multifunctional, highly integrated, and general optical intelligent systems. This work introduces an all‐optical autoencoder (OAE) framework utilizing the bidirectional multiplexing property of diffractive deep neural networks (D2NN). The OAE encodes and decodes input wavefields in the diffractive latent space (DLS), enhancing information mining and multifunctional integration. Applied to image denoising, noise‐resistant classification, and image generation, the framework leverages bidirectional processing for efficient optical generative models and integrated systems.
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1863-8880
1863-8899
DOI:10.1002/lpor.202401945