Image mean learning circuit based on memristive cross architecture
A resistance and mean value technology, applied in the field of picture mean value learning circuits, can solve the problems of inconvenient operation of changing series resistance, increasing the resistance value of series resistance, and difficulty in realizing large-scale reading and writing of memristive crossover architecture, and achieves mathematical models and calculations. The effect of reducing the accuracy requirements, high tolerance, and reducing the time required
Inactive Publication Date: 2017-08-08
SOUTHWEST UNIVERSITY
View PDF5 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
In this case, the final write resistance of the memristor is controlled by the series resistance, but the half-selected memristors (memristors in the same row or column as the target memristor) on the crossover structure in the actual circuit will change the phase increase the resistance of the series resistor, resulting in write noise
Moreover, due to the inconvenient operation of changing the series resistance, it is difficult to realize the large-scale reading and writing of the memristive cross-architecture.
Therefore, the use of series resistors to achieve multi-level storage still has great limitations.
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0021] The preferred embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings; it should be understood that the preferred embodiments are only for illustrating the present invention, rather than limiting the protection scope of the present invention.
[0022] Such as figure 1 In the multi-layer MCA structure shown, the first layer of MCA is called memory memristive cross architecture (MMCA), and the second layer is called impact factor memristive cross architecture (IMCA). The two layers are connected through CMOS unit 1 . MMCA is used to store pictures, memristor is called storage factor (for color pictures, construct three MMCA, corresponding to RGB three channels), initial value is randomly generated, IMCA is used to mark pixels, memristor 3 is called influence factor, The initial values are all set to intermediate values. When learning pictures, the CMOS unit will read and write to the two layers of MCA at the...
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
Login to View More Abstract
The invention discloses a picture mean value learning circuit based on a memristor cross structure, which is characterized in that it includes an MMCA layer composed of a memristor cross structure and an IMCA layer composed of a memristor cross structure, the MMCA layer and the IMCA The layers are connected by a CMOS unit; the CMOS unit adjusts the resistance value of the memristor in the MMCA layer according to the difference between the mean value of the resistance value corresponding to the input pixel and the mean value of the stored resistance value in the MMCA layer, so that the MMCA layer The resistance value stored in is close to the resistance value corresponding to the average value of the input pixel; the CMOS unit adjusts the resistance value of the memristor in the IMCA layer according to the error between the resistance value corresponding to the input pixel and the resistance value stored in the MMCA layer. value, when the error value is less than a certain threshold, the resistance value of the memristor in the IMCA layer increases, otherwise the resistance value of the memristor decreases. The invention proposes to use the average value method to learn and write pictures to the memristive cross-frame, and has the characteristics of high fault tolerance and parallel processing.
Description
technical field [0001] The invention relates to a neural circuit, in particular to a picture mean value learning circuit based on a memristive cross-frame. Background technique [0002] A memristor is a two-terminal element with a memory function, and its resistance is controlled by the strength, polarity, and duration of the power supply. After power off, it is able to maintain the resistance value at power off. As a nanoscale device, memristor has an important application in memory based on memristive cross-architecture (MCA). Before the emergence of memristors, the crossover architecture usually uses switches to store data 0 and 1 at the crosspoints, but after the emergence of memristors, it can not only store data 0 and 1 with high resistance and low resistance to realize the switch function, but also Multi-level storage can be realized. Memory devices using memristors as storage units are called non-volatile resistive random access memory (RRAM), which is different f...
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
Login to View More IPC IPC(8): H03K19/0948
Inventor 李传东何兴陈玲黄廷文
Owner SOUTHWEST UNIVERSITY