Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

SNN workload prediction method and system

A workload and prediction method technology, applied in the field of spiking neural networks, can solve problems such as poor applicability of random connectivity networks, low communication load prediction accuracy, and deeper understanding of SNN workloads, etc., to ensure high-performance operation Effect

Inactive Publication Date: 2022-02-25
JIANGNAN UNIV
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in actual operation, SWAM has the problems of low prediction accuracy for communication load, complicated process and poor applicability for random connectivity networks.
The reason is that there is no deeper understanding of the SNN workload

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
  • SNN workload prediction method and system
  • SNN workload prediction method and system
  • SNN workload prediction method and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0052] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, so that those skilled in the art can better understand the present invention and implement it, but the examples given are not intended to limit the present invention.

[0053] like figure 1 Shown, be the SNN workload prediction method in the preferred embodiment of the present invention, it comprises the following steps:

[0054] S1, build SNN workload model based on NEST emulator, described SNN workload model comprises: memory load model, computation load model and communication load model; Specifically:

[0055] In the memory load model, the total memory consumption M of each process includes M 0 , M n and M s Three parts, as shown in the following formula:

[0056]

[0057] Among them, M 0 is the basic memory consumption required for NEST to run, neuron memory overhead M n and the synaptic memory overhead M s As shown in the followin...

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 an SNN workload prediction method and system. The method comprises the steps: the following steps: S1, building an SNN workload model based on an NEST simulator, and the SNN workload model comprises a memory load model, a calculation load model and a communication load model; s2, collecting parameters of the SNN workload model, wherein the parameters comprise memory parameters, time parameters and network parameters; and S3, constructing a load calculation function according to the SNN workload model, processing the obtained parameters, and predicting the workload of the SNN target network under a plurality of nodes. According to the SNN workload prediction method and system, the problem of reasonable matching between the SNN workload and the calculation platform can be solved, the mapping result of the SNN network on the calculation platform can be accurately predicted, mapping guidance is provided for the calculation platform on the basis, and high-performance operation of the platform is ensured in a mode of reasonably distributing the calculation nodes.

Description

technical field [0001] The invention relates to the technical field of spiking neural networks, in particular to a SNN workload prediction method and system. Background technique [0002] Brain-like computing represented by the third-generation artificial neural network - Spike neuron network (SNN) has been widely used in the field of neuroscience. SNN itself has obvious distributed computing characteristics, and only a certain scale of SNN can show a strong level of intelligence. Therefore, building a large-scale distributed cluster is the mainstream way to form a brain-inspired computing platform. [0003] For distributed brain-inspired computing platforms, the mismatch between SNN workloads and computing platforms will cause the computing energy efficiency of special-purpose brain-inspired systems to be even inferior to general-purpose computer systems. Therefore, realizing a reasonable match between SNN workload and computing platform is one of the key issues in the cur...

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): G06F9/50G06N3/04G06N3/063
CPCG06F9/505G06N3/049G06N3/065
Inventor 柴志雷华夏吴秦刘登峰陈璟肖志勇
Owner JIANGNAN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com