Build Machine Learning Models Using Microsoft Azure - Arpita - CCDays
This session guides through the process followed for development of machine learning model on Microsoft Azure from creation of machine learning model to deployment.
Presented as part of Cloud Community Days on 18th June
Agenda
NEED OF CLOUD
SERVICEFOR MACHINE
LEARNING
MACHINE LEARNING
SERVICES OFFERED BY
MICROSOFT AZURE
PIPELINE FOR
DESIGNING A MACHINE
LEARNING MODEL
STEPS TO ACCESS
MICROSOFT AZURE ML
RESOURCES
BUILDING MACHINE
LEARNING MODEL
USING AZURE
MACHINE LEARNING
MODEL DEPLOYMENT
4.
Need of cloudservice for machine learning
Bottlenecks with traditional resources:
Computing time
Storage issue
Solutions offered by cloud services:
Access to fast computing hardware
Containers for storage
5.
Machine learning Serviceby
Microsoft Azure
Fully Managed
Integrated
Easy Deployment
Best Collection of ML
algorithms
6.
Pipeline for designingmachine learning model
Understanding the problem
Data Acquisition
Data cleaning and feature selection
Train the model (ML Algorithm implementation)
Test and validate the model
Deploy the model
Building machine learningmodel Using Azure
10
Create target for
computation
Attach the target to
workspace
Set dependencies for
training
Create experiment
Start run process
Complete the run process
11.
Machine Learning ModelDeployment
• Process flow:
• Files needed:
• Script to initiate and run the model
• Dependencies used
• Configuration file for deployment
Register the model Prepare to deploy Test the model
Deploy to compute
target