US 11,600,060 B2Granted

Nonlinear all-optical deep-learning system and method with multistage space-frequency domain modulation

AssigneeTsinghua University

InventorsQionghai Dai, Tao Yan, Jiamin Wu, Xing Lin

Life of this patent

Estimating the term…

Loading patent…

Abstract

The present disclosure discloses a nonlinear all-optical deep-learning system and method with multistage space-frequency domain modulation. The system includes an optical input module, configured to convert input information to optical information, a multistage space-frequency domain modulation module, configured to perform multistage space-frequency domain modulation on the optical information generated by the optical input module so as to generate modulated optical information, and an information acquisition module, configured to transform the modulated optical information onto a Fourier plane or an image plane, and to acquire the transformed optical information so as to generate processed optical information.

Claims17

Dependent claims

A nonlinear all-optical deep-learning system with multistage space-frequency domain modulation, comprising:

an optical input module, configured to convert input information to optical information;
a multistage space-frequency domain modulation module, configured to perform multistage space-frequency domain modulation on the optical information generated by the optical input module so as to generate modulated optical information;
and
an information acquisition module, configured to transform the modulated optical information onto a Fourier plane or an image plane, and to acquire the transformed optical information so as to generate processed optical information;
wherein the multistage space-frequency domain modulation module comprises:
one or more frequency-domain modulation module, configured to perform frequency-domain modulation on optical information in Fourier space;
and
one or more spatial-domain modulation module, configured to perform spatial-domain modulation on optical information in spatial domain,
wherein, the one or more frequency-domain modulation module and/or the one or more spatial-domain modulation module are cascaded alternately;
wherein the modulation module at each odd-numbered stage of the multistage space-frequency domain modulation module is the frequency-domain modulation module, and the modulation module at each even-numbered stage of the multistage space-frequency domain modulation module is the spatial-domain modulation module;
wherein, when a total number of the modulation modules in the multistage space-frequency domain modulation module is an odd number, the last stage of the multistage space-frequency domain modulation module is the frequency-domain modulation module;
and
when the total number of the modulation modules in the multistage space-frequency domain modulation module is an even number, the last stage of the multistage space-frequency domain modulation module is the spatial-domain modulation module.

10 more elements

A nonlinear all-optical deep-learning method with multistage space-frequency domain modulation, comprising:

establishing a numerical simulation model of an optical system with multistage space-frequency domain modulation, the system comprising an optical input module configured to convert input information to optical information, a multistage space-frequency domain modulation module configured to perform multistage space-frequency domain modulation on the optical information generated by the optical input module so as to generate modulated optical information, and an information acquisition module configured to transform the modulated optical information onto a Fourier plane or an image plane and to acquire the transformed optical information so as to generate processed optical information;
and
manufacturing a physical model of the optical system in which a hardware system is installed to implement a target function,
wherein establishing the numerical simulation model of the optical system with multistage space-frequency domain modulation comprises:
setting a wavelength of a light source in the optical input module;
obtaining normalized gray values of an image of an input object as the intensity of the light inputted to the multistage space-frequency domain modulation module, by irradiating the input object uniformly with light emitted from the light source;
and
modulating the inputted light through multistage space-frequency domain modulation module and acquiring it in the information acquisition module.

5 more elements

Dependencies are inferred from the claim text.

Facts

Application: 16/893,071Filed 4 Jun 2020
Priority: CN 201910487242.05 Jun 2019
Published: US 2020/0387780 A110 Dec 2020
Granted: 7 Mar 2023
Term ends: Not estimatedThe source records don’t support an estimate.

Classification11 codes

G06V 10/82
Using pattern recognition or machine learning: using neural networks
G06N 3/04
Neural networks: architecture, e.g. interconnection topology
G06N 3/0464
Architecture, e.g. interconnection topology: convolutional networks [CNN, ConvNet]
G06N 3/0675
Using optical means: using electro-optical, acousto-optical or opto-electronic means