Search
Menu

All-Optical Neural Network Uses Parallel Computation to Speed Problem-Solving

Facebook X LinkedIn Email
In a step toward making the use of large-scale optical neural networks practical, researchers at The Hong Kong University of Science and Technology have demonstrated a multilayer all-optical artificial neural network.

In conventional hybrid optical neural networks, optical components are typically used for linear operations, while nonlinear activation functions — the functions that simulate the way neurons in the human brain respond — are implemented electronically, because nonlinear optics usually require high-power lasers that are difficult to implement in an optical neural network.

Researchers demonstrated the first two-layer, all-optical artificial neural network with nonlinear activation functions. These types of functions are required to perform complex tasks such as pattern recognition. Courtesy of Olivia Wang, Peng Cheng Laboratory.
Researchers demonstrated the first two-layer, all-optical artificial neural network with nonlinear activation functions. These types of functions are required to perform complex tasks such as pattern recognition. Courtesy of Olivia Wang, Peng Cheng Laboratory.

The researchers built and tested an all-optical neural network in which linear operations were programmed by spatial light modulators and Fourier lenses, while nonlinear optical activation functions were realized using laser-cooled atoms with electromagnetically induced transparency.

“This light-induced effect can be achieved with very weak laser power,” professor Shengwang Du said. Because this effect is based on nonlinear quantum interference, the researchers surmised that it could be possible to extend their system into a quantum neural network that would be able to solve problems that could not be solved by classical methods.


To confirm the feasibility of their approach, the researchers constructed a two-layer, fully connected, all-optical neural network with 16 inputs and two outputs. The team used its all-optical network to classify the order and disorder phases of the Ising model, a statistical model of magnetism. The results showed that the all-optical neural network was as accurate as a well-trained computer-based neural network.

The researchers said that the hardware system for their all-optical network is reconfigurable for different applications without the need to modify the physical structure. They plan to expand their all-optical approach to large-scale all-optical deep neural networks with complex architectures designed for specific applications such as image recognition. This next step will help demonstrate that the use of parallel computation for all-optical neural networks can work at larger scales.

“Although our work is a proof-of-principle demonstration, it shows that it may become possible in the future to develop optical versions of artificial intelligence,” Du said.

“Our all-optical scheme could enable a neural network that performs optical parallel computation at the speed of light while consuming little energy,” professor Junwei Liu said. “Large-scale, all-optical neural networks could be used for applications ranging from image recognition to scientific research.”

The research was published in Optica, a publication of OSA, The Optical Society (https://doi.org/10.1364/OPTICA.6.001132).   

Published: August 2019
Glossary
nonlinear optics
Nonlinear optics is a branch of optics that studies the optical phenomena that occur when intense light interacts with a material and induces nonlinear responses. In contrast to linear optics, where the response of a material is directly proportional to the intensity of the incident light, nonlinear optics involves optical effects that are not linearly dependent on the input light intensity. These nonlinear effects become significant at high light intensities, such as those produced by...
deep learning
Deep learning is a subset of machine learning that involves the use of artificial neural networks to model and solve complex problems. The term "deep" in deep learning refers to the use of deep neural networks, which are neural networks with multiple layers (deep architectures). These networks, often called deep neural networks or deep neural architectures, have the ability to automatically learn hierarchical representations of data. Key concepts and components of deep learning include: ...
artificial intelligence
The ability of a machine to perform certain complex functions normally associated with human intelligence, such as judgment, pattern recognition, understanding, learning, planning, and problem solving.
Research & TechnologyeducationAsia-PacificThe Hong Kong University of Science and TechnologyOpticsLasersnonlinear opticsdeep learningneural networksall-optical neural networksImagingimage recognitionartificial intelligenceparallel computationmultilayer all-optical neural networkTech Pulse

We use cookies to improve user experience and analyze our website traffic as stated in our Privacy Policy. By using this website, you agree to the use of cookies unless you have disabled them.