Method and device for achieving all-optical nonlinear activation function of optical neural network

A neural network and activation function technology, applied in the field of optical neural networks, can solve the problems of high requirements for transimpedance amplifiers and high power consumption, and achieve the effect of improving the overall operation speed

Active Publication Date: 2020-10-30
UNITED MICROELECTRONICS CENT CO LTD
View PDF4 Cites 25 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, these methods have high requirements for the transimpedance amplifier (TIA) and consume a lot of power

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Method and device for achieving all-optical nonlinear activation function of optical neural network
  • Method and device for achieving all-optical nonlinear activation function of optical neural network
  • Method and device for achieving all-optical nonlinear activation function of optical neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment approach

[0041] According to the first embodiment of the present invention, a method for realizing an all-optical nonlinear activation function of an optical neural network is provided. see Figure 1-Figure 5 , Figure 9 , figure 2 It is a structural schematic diagram of an all-optical nonlinear activation function unit according to an embodiment of the present invention, including a 2×2 Mach-Zehnder interferometer and two phase shifters (the first phase shifter φ, the second phase shifter θ ), the 2×2 Mach-Zehnder interferometer consists of two 2×2 beam splitters, the phase shifter φ is set on the signal arm of the Mach-Zehnder interferometer, and the phase shifter θ is set on the Mach-Zehnder interferometer on the upper interference arm in the instrument. figure 2 The phase shifter φ in corresponds to figure 1 The first phase shifting module in . figure 2 The two beam splitters (the first beam splitter and the second beam splitter) and the phase shifter θ in are corresponding...

no. 2 approach

[0085] According to the second embodiment of the present invention, an apparatus for realizing an all-optical nonlinear activation function of an optical neural network is provided. In one example, see figure 1 , figure 2 and Figure 9 , the all-optical nonlinear activation function implementation device of the optical neural network includes a nonlinear activation function unit, and the nonlinear activation function unit includes a Mach-Zehnder interferometer and a first phase shifter φ; the Mach-Zehnder interferometer consists of two 2×2 beam splitter and a second phase shifter θ, the second phase shifter θ is set on the upper interference arm of the Mach-Zehnder interferometer, and the first phase shifter φ is set on the Mach-Zehnder interferometer on the signal arm of the instrument. For the further functions, working process, and technical effects of the all-optical nonlinear activation function realization device of the optical neural network in this embodiment, plea...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a method and a device for achieving an all-optical nonlinear activation function of an optical neural network. The method comprises the following steps: acquiring a to-be-processed signal optical signal and a reference optical signal coherent with the to-be-processed signal optical signal; inputting the to-be-processed signal optical signal and the reference optical signalinto a first phase shift module, wherein the first phase shift module performs phase shift operation on the to-be-processed signal optical signal and the reference optical signal to obtain an opticalsignal of a first array; and inputting the optical signal of the first array into an optical interference module, performing nonlinear operation on the optical signal of the first array in the opticalinterference module to obtain an optical signal of a second array, and outputting the optical signal of the second array as a nonlinear response of the signal optical signal to be processed. According to the technical scheme, the problem that an existing optical nonlinear function calculation unit has high requirements for optical power and a transimpedance amplifier is solved, parameters are adjustable, and a provided nonlinear function is flexible and controllable.

Description

technical field [0001] The invention relates to the technical field of optical neural networks, in particular to a method and device for realizing an all-optical nonlinear activation function. Background technique [0002] In the most popular deep learning of artificial intelligence, its operation process mainly involves two parts: matrix multiplication and nonlinear activation function. Nonlinear computing is the root of the strong expressive ability of artificial neural networks. It can speed up the convergence of the network and improve the accuracy of recognition. It is an indispensable part of the neural network. [0003] The essence of using light to complete deep learning algorithms is to use light to realize matrix multiplication and nonlinear activation functions. To use light to complete the matrix multiplication operation, the existing method is to use Mach-Zehnder interferometer (MZI) array, micro-ring resonator (MRR) array, etc. to realize. However, there are ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06N3/067G06N3/04G06F17/16G02F1/21
CPCG06N3/0675G06F17/16G02F1/21G06N3/045
Inventor 李强田野刘胜平赵洋王玮冯俊波郭进
Owner UNITED MICROELECTRONICS CENT CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap