Programmable matrix operation with reconfigurable time-wavelength plane manipulation and dispersed time delay

Huang, Yuyao; Zhang, Wenjia; Yang, Fan; Du, Jiangbing; He, Zuyuan

doi:10.1364/oe.27.020456

Cited by 29 publications

(20 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Because of the line-by-line spectral manipulation capability of the SLMs, any linear signal processing function can be potentially executed. Typical examples include correlation [118], spatial Fourier transform [119], [120], matrix calculation [121] and mode shaping [122], [123].…”

Section: Broadband Analog Signal Processingmentioning

confidence: 99%

Microwave Photonic Radars

Pan

Zhang

2020

J. Lightwave Technol.

279

View full text Add to dashboard Cite

As the only method for all-weather, all-time and longdistance target detection and recognition, radar has been intensively studied since it was invented, and is considered as an essential sensor for future intelligent society. In the past few decades, great efforts were devoted to improving radar's functionality, precision, and response time, of which the key is to generate, control and process a wideband signal with high speed. Thanks to the broad bandwidth, flat response, low loss transmission, multidimensional multiplexing, ultrafast analog signal processing and electromagnetic interference immunity provided by modern photonics, implementation of the radar in the optical domain can achieve better performance in terms of resolution, coverage, and speed which would be difficult (if not impossible) to implement using traditional, even state-of-the-art electronics. In this tutorial, we overview the distinct features of microwave photonics and some key microwave photonic technologies that are currently known to be attractive for radars. System architectures and their performance that may interest the radar society are emphasized. Emerging technologies in this area and possible future research directions are discussed.

show abstract

Section: Broadband Analog Signal Processingmentioning

confidence: 99%

Microwave Photonic Radars

Pan

Zhang

2020

J. Lightwave Technol.

279

View full text Add to dashboard Cite

show abstract

“…Carrying out computations in the complete real-value domain with high numerical accuracy is the basic requirement for regression, which is still challenging for the existing ONN chips. Firstly, for non-coherent ONN architectures 16 , 20 – 22 , input values are represented by non-negative optical intensities, causing incompleteness of the numerical domain. In contrast, coherent ONN architectures 15 , 28 , 29 adopt optical fields to represent real-valued inputs and homodyne detection to yield real-valued outputs, showing the capability of computing in the complex-valued domain.…”

Section: Introductionmentioning

confidence: 99%

“…To solve the problem, optical implementations of neural network (ONNs) have been recently proposed and demonstrated to realize high-speed and energyefficient AI hardware [10][11][12][13]. Linear propagation of light equivalently carries out the linear computing of ANNs [14][15][16][17][18]; ultra-wide optical transparent spectrum and high-speed modulators/detectors enable a fast clock rate (tens of GHz) [19,20]; and non-volatile photonic memory makes the computing 'zero-consuming' [21].…”

Section: Introductionmentioning

confidence: 99%

Optical coherent dot-product chip for sophisticated deep learning regression

Xu¹,

Wang²,

Shu

et al. 2021

Light Sci Appl

View full text Add to dashboard Cite

Optical implementations of neural networks (ONNs) herald the next-generation high-speed and energy-efficient deep learning computing by harnessing the technical advantages of large bandwidth and high parallelism of optics. However, due to the problems of the incomplete numerical domain, limited hardware scale, or inadequate numerical accuracy, the majority of existing ONNs were studied for basic classification tasks. Given that regression is a fundamental form of deep learning and accounts for a large part of current artificial intelligence applications, it is necessary to master deep learning regression for further development and deployment of ONNs. Here, we demonstrate a silicon-based optical coherent dot-product chip (OCDC) capable of completing deep learning regression tasks. The OCDC adopts optical fields to carry out operations in the complete real-value domain instead of in only the positive domain. Via reusing, a single chip conducts matrix multiplications and convolutions in neural networks of any complexity. Also, hardware deviations are compensated via in-situ backpropagation control provided the simplicity of chip architecture. Therefore, the OCDC meets the requirements for sophisticated regression tasks and we successfully demonstrate a representative neural network, the AUTOMAP (a cutting-edge neural network model for image reconstruction). The quality of reconstructed images by the OCDC and a 32-bit digital computer is comparable. To the best of our knowledge, there is no precedent of performing such state-of-the-art regression tasks on ONN chips. It is anticipated that the OCDC can promote the novel accomplishment of ONNs in modern AI applications including autonomous driving, natural language processing, and scientific study.

show abstract

“…Therefore, both higher computing speed and lower energy cost can be obtained once adopting the up-to-date technology of optoelectronics. Time-stretch method [17], which is often used in fields of signal processing [18][19][20][21] and imaging [22][23][24][25], provides a way of manipulating pulses in frequency-time plane. These pulses which have been broadened by adopting time-stretch method can be served as data carriers for serial vector computations once they were modulated with encoded data.…”

Section: Introductionmentioning

confidence: 99%

Optoelectronic convolutional neural networks based on time-stretch method

Zang

Chen

Yang

et al. 2021

Sci. China Inf. Sci.

View full text Add to dashboard Cite

In this paper, a new architecture of optoelectronic convolutional neural networks (CNNs) based on time-stretch method is proposed. In this loop-shaped structure mainly composed of fiber optical and electronic devices, computations of data from each layer of CNN which are carried by light pulses with high repetition rate can be accomplished in a serial way. Therefore, a 5-layer CNN with two convolution layers, two mean pooling layers and one fully-connected layer are implemented. Under the test of handwriting digit recognition, its accuracy can reach up to 95% under ideal circumstances. Tests under different relative noise levels have been conducted and analyzed as well.

show abstract

Programmable matrix operation with reconfigurable time-wavelength plane manipulation and dispersed time delay

Cited by 29 publications

References 23 publications

Microwave Photonic Radars

Microwave Photonic Radars

Optical coherent dot-product chip for sophisticated deep learning regression

Optoelectronic convolutional neural networks based on time-stretch method

Contact Info

Product

Resources

About