Pulse neural network simulation strategy based on GPU

A technology of pulse neural network and pulse, which is applied in the field of pulse neural network and high-performance computing, can solve the problems of unresearchable parallel algorithm, inability to make full use of GPU, and inability to make full use of pulse sparsity, etc., to achieve strong versatility and scalability performance, low latency, and low energy consumption

Pending Publication Date: 2022-05-31
NANJING UNIV OF AERONAUTICS & ASTRONAUTICS
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the GPU-based spiking neural network model still has the following problems. First, the existing technology cannot make full use of GPU resources such as memory and threads. Second, the existing simulation strategy cannot fully utilize the advantages of spiking sparsity. The parallelization algorithm of

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
  • Pulse neural network simulation strategy based on GPU
  • Pulse neural network simulation strategy based on GPU
  • Pulse neural network simulation strategy based on GPU

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0044] Below in conjunction with example the present invention will be further described.

[0045] Such as figure 1 As shown, the present invention mainly includes the following steps, and the specific steps of the step S1 are:

[0046] S11. Create m neurons, number the neurons sequentially from 0 to m-1, and each neuron has a corresponding unique number id. And initialize the membrane voltage V of each neuron mem , threshold voltage V thr , resting voltage V rest , time constant τ m , the resistance constant R m , connect the corresponding neurons according to the connectivity rate conn;

[0047]S12. Then sequentially number the connection synapses between neurons, starting from neuron 0, sequentially number the synapses where the source neuron is the current neuron, and initialize the delay parameter delay and weight w for it.

[0048] Further, the specific steps of the step S2 are:

[0049] S21. Create an array to store the membrane voltage V of m neurons mem , thr...

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 pulse neural network simulation strategy based on a GPU. The pulse neural network simulation strategy comprises the following steps: initializing a neural network structure and a network weight; loading parameters and a network structure in the GPU, and creating a pulse queue; calculating a neuron membrane voltage according to the pulse distribution condition in the pulse queue; according to the membrane voltage value of the neuron and a threshold value, whether a pulse is emitted is judged; and the process from S3 to S5 is repeated until iteration is completed. According to the method, the simulation speed of the spiking neural network is accelerated, the advantages of GPU parallel computing and the characteristics of sparsity and concurrency of the spiking neural network are brought into full play, and meanwhile simulation of a larger-scale spiking neural network model is supported.

Description

technical field [0001] The invention belongs to the fields of pulse neural network and high-performance computing, and in particular relates to the design and realization of a GPU-based pulse neural network simulation strategy. Background technique [0002] With the rapid growth of labeled data and computing power, deep learning has been widely used in many fields, but training larger networks means more data, faster computing efficiency and higher energy consumption. In contrast, the human brain not only has a high level of intelligence but also consumes only about 25 watts of power. Therefore, as a new type of neural network, the bionic network represented by spiking neural network, which replaces real-valued input with discrete sequences, has received more and more attention. [0003] From the perspective of bionics, the spiking neural network is more biologically interpretable because it operates in the form of real biological tissue. It refers to the biological learni...

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
IPC IPC(8): G06N3/04G06N3/10
CPCG06N3/049G06N3/10Y02D10/00
Inventor 袁家斌夏涛
Owner NANJING UNIV OF AERONAUTICS & ASTRONAUTICS
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