Unlock instant, AI-driven research and patent intelligence for your innovation.

Method for realizing calculation of endosensory reserve pool based on photosynaptic device

A reserve pool and device technology, applied in the field of brain-like neural network algorithms, can solve the problem that the calculation of the reserve pool in the sensor has not yet been realized, and achieve the effect of saving hardware costs and reducing delay

Pending Publication Date: 2022-07-08
ZHEJIANG LAB +1
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although many photo-synaptic devices have been reported, the internal storage pool computing for sensing based on self-powered photo-synaptic devices has not been realized so far.

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 for realizing calculation of endosensory reserve pool based on photosynaptic device
  • Method for realizing calculation of endosensory reserve pool based on photosynaptic device
  • Method for realizing calculation of endosensory reserve pool based on photosynaptic device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0061] In Example 1, the method for calculating the storage pool in the sensor of the present invention is directly applied to image processing. The method of the present invention is further described by taking the recognition of four grayscale face images of 28 columns*35 rows of pixels as an example. figure 1 In (a) is the schematic diagram of the dynamic reserve pool calculation, that is, the instantaneous input of the photosynaptic reserve device at time t constitutes the input state u(t), with the photocurrent or photoconductance of the photosynaptic reserve device x ( t), which maps the input u(t) to a high-dimensional feature space y(t). figure 1 (b) in (b) is a schematic diagram of the process of gray-scale face recognition from input to readout neural network. First select four target faces and preprocess them to 28×35 pixels, such as figure 1 The face image in (c) is shown. Each pixel value ranges from 0 to 5 to reflect their gray level. The value of pure white ...

Embodiment 2

[0065] In Embodiment 2 of the present invention, the in-sensor storage pool calculation of the present invention is directly applied to the processing of time-series video signals. like Figure 4 As shown in (a), a 5 × 5 photo-synaptic reservoir device array composed of a reservoir layer is used to map and collect the motion information of cars at the intersection. Different car orientations naturally generate corresponding time-series optical signals input to the reservoir layer, recording the photocurrent response of the 5×5 photo-synaptic reservoir device. with Cs 2 AgBiBr 6 As an example of a photo-synaptic device, due to the SNDP and SRDP effects of the photo-synaptic device, that is, the photocurrent or photoconductive amplitude of the photo-synaptic reservoir device produces a nonlinearly coupled photo-electric response to the time-series optical signal, different optical pulse sequences will cause Cs 2 AgBiBr 6 Different photocurrent responses of optosynaptic devi...

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 for realizing intra-sensory reserve pool calculation based on a photosynaptic device, which comprises the following steps of: converting a space pixel signal or a time sequence signal into a time sequence optical signal, inputting the time sequence optical signal into a photoelectric device, mapping original space information or original time sequence information to a high-dimensional space by utilizing nonlinear coupling of the photoelectric device to the time sequence optical signal response, and calculating the intra-sensory reserve pool based on the photosynaptic device. And then features of original input signals are effectively extracted from high-dimensional information by using a simple matrix multiplication operation of the memristor array. The intrasensory reserve pool calculation based on the photosynaptic device not only can process a static image classification problem, but also can process dynamic video information. Compared with an existing visual computing technology based on a pure resistance reserve pool which needs to be serially connected with an additional photoelectric sensing device, the optical signal can be directly received through reserve pool computing based on the optical synapse device, parallel sensing memory computing is achieved, delay is reduced, and meanwhile hardware cost is greatly saved.

Description

technical field [0001] The invention relates to the technical field of brain-like neural network algorithms, in particular to a method for realizing intrasensory reserve pool calculation based on an optical synapse device. Background technique [0002] In the existing traditional computing units based on the von Neumann architecture, the physical separation of sensing, memory and processing units produces a lot of energy consumption and time delay in the process of matrix multiplication, which seriously limits its use in unmanned applications. Applications in real-time information processing scenarios such as driving, robotics, and industrial manufacturing. In addition, the rapid development of the Internet of Things and the increasing demand for image resolution have brought about the explosive growth of visual information, and it has become particularly urgent to solve the huge energy consumption problem associated with the complex visual process. Self-powered photoelectr...

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): G06N3/04G06N3/067
CPCG06N3/049G06N3/067Y02E60/00
Inventor 田博博劳婕闫梦阁彭晖段纯刚
Owner ZHEJIANG LAB