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

Machine learning model framework development method and system based on containerization technology

A technology of machine learning model and containerization technology, applied in neural learning methods, computer simulation, biological neural network models, etc., can solve problems such as difficult to find software architecture

Pending Publication Date: 2021-04-09
SHENZHEN JINGTAI TECH CO LTD
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, when users need to build a similar environment on a host in an offline environment or a self-built cloud server, it is difficult to find a best-practice software architecture that can satisfy

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
  • Machine learning model framework development method and system based on containerization technology
  • Machine learning model framework development method and system based on containerization technology
  • Machine learning model framework development method and system based on containerization technology

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0040] Take developing a machine learning based image recognition program as an example.

[0041] A machine learning algorithm development department can build a model based on the literature algorithm, build an image recognition program through a selected machine learning framework and the algorithm implementation code on the framework, and train and verify the model through limited test data , if the verification result is good, you can submit the code of the model to the code warehouse (such as Github).

[0042] The continuous integration service (such as Jenkins) recognizes the change of the code warehouse, which triggers the container image construction process, and builds a base image belonging to the current code change. At the same time, the built image can be selected from the Binder page and deployed as an input to deploy a container instance. The startup process of the instance will combine the machine learning framework, the code of the developer's custom model, an...

Embodiment 2

[0048] Take the example of developing a molecular generation algorithm for drug design.

[0049] Artificial intelligence engineers can construct molecular generation models based on convolutional neural networks (RNN) based on literature, and push the corresponding codes to code warehouses (such as Bitbucket) after completing model construction through open source machine learning development frameworks. Generally speaking, the model parameters of this type of neural network, such as learning rate (learning rate), batch size (batch size), etc., require the model to be continuously controlled during the specific debugging and training process in order to obtain better results.

[0050] After the code is pushed, the continuous integration service (Jenkins) detects the changes in the code warehouse and starts the corresponding container image construction. The image construction and deployment as a container instance can be realized by Binder. The instance startup process include...

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 provides a machine learning model framework development method and system based on a containerization technology, and the method comprises the following steps: machine learning model development: building a corresponding machine learning model, building a model in a programming mode, and submitting a corresponding code to a code warehouse; model interactive testing: selecting proper test data to verify the developed machine learning model; model parameter optimization and monitoring: directly calling the successfully debugged and verified machine learning model and parameters by other users, achieving optimal machine learning model training and result monitoring by parameter adjustment on the code level, and finally, acquiring a machine learning model and parameters for deployment. According to the invention, the cross-platform extensible deployment implementation capability is realized through the containerization technology.

Description

technical field [0001] The invention relates to an auxiliary method for drug research and development, in particular to a method and system based on a containerized technology-based machine learning model framework. Background technique [0002] There are many development frameworks for machine learning, among which neural network-based development frameworks include Tensorflow, Microsoft CNTK, Theano, Caffe, Keras, Torch, Sci-kit Learn, MXNet, etc. Choose an appropriate framework for specific machine learning development work. Generally speaking, after developers model the model based on specific test data and business problems, they package and release the code of the model and the dependent machine learning framework, which can be used by others in specific business Test on real data in the scenario. Different from other software functional algorithm tests, machine learning algorithm testing is often a process of testing and debugging. The parameters of the model are adj...

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): G06N3/10G06N3/08G06F8/71G06F8/72G06F11/36
CPCG06N3/105G06N3/08G06F8/71G06F8/72G06F11/3688
Inventor 吴楚楠温书豪徐旻范陕珊张佩宇刘阳马健赖力鹏
Owner SHENZHEN JINGTAI TECH CO LTD
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