A voltage-type neural network based on polystate memristor and its operation method
A neural network and memristor technology, applied in the field of neural networks, can solve the problems of high energy consumption, high resource allocation requirements, and high hardware costs of artificial intelligence systems, and achieve the effect of improving computing efficiency and stability
Active Publication Date: 2021-10-26
NANJING UNIV OF POSTS & TELECOMM
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
In this case, while the neural network-based intelligent system takes advantage of its advantages, it also exposes the defects such as high energy consumption of the artificial intelligence system, high hardware cost, and high resource allocation requirements.
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
[0039] The present invention will be further described below in conjunction with the accompanying drawings.
[0040] A voltage-type neural network based on a polystate memristor, including a signal input module, a weight network module, a polarity register module, a Sum summing unit, a bias register module and an activation function module.
[0041] Signal input modules such as Figure 1-Figure 2 Shown, including the absolute value circuit part and the polar characteristic signal part. The absolute value circuit part adopts the absolute value circuit with high input impedance, and the polar characteristic signal part is a functional circuit with rail-to-rail operational amplifier as the core. V pluse When it is a positive polarity level, the output modulus signal |V pluse | and polarity signal pulse V h , when the input signal V pluse When it is a negative polarity level, the output modulus signal |V pluse | and polarity signal pulse V l .
[0042] The signal input mod...
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 voltage-type neural network based on a polymorphic memristor and an operation method thereof, and proposes a voltage-type neuron circuit constructed by a brain-like device memristor combined with traditional devices, which can simulate and realize forward The neural network operation adopts the weight mode with the memristive device as the core, which effectively reduces the resource consumption of storage and operation in the neural network operation. Combined with the principles of other types of electronic devices such as MOS transistors, low-power op amps, rail-to-rail op amp technology, and digital circuits, the present invention specifically solves the problem of memristors being designed as core devices in neural networks. Signal input, weight network, cumulative summation, and activation level design issues, realize the processing of positive and negative signals in the multiplier and the transfer between neural network layers, and build the corresponding weight matrix model and neuron network circuits.
Description
technical field [0001] The invention relates to the technical field of neural networks, and mainly relates to a voltage-type neural network system based on polymorphic memristors. Background technique [0002] Regarding the design of the neural network model, the existing technology has realized some functions similar to the biological neural network, established a nonlinear system similar to the neural network, and realized the intelligence of the system, but most of them still use large-scale FPGA and Hardware programming design represented by computer (GPU, TPU), that is, system design based on transistor 01 model. In this case, while the neural network-based intelligent system takes advantage of its advantages, it also exposes the shortcomings of artificial intelligence systems such as extremely high energy consumption, high hardware costs, and high resource allocation requirements. [0003] Compared with the biological neural network system composed of electronic syste...
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 Patent Type & Authority Patents(China)
IPC IPC(8): G06N3/063
CPCG06N3/063
Inventor 肖建张粮张子恒洪聪童祎张翼郭宇锋
Owner NANJING UNIV OF POSTS & TELECOMM