Understanding End-to-End Machine Learning Process (Part 5 of 5)

By Ashwin Venugopal -

 





To read part 1, please click here
To read part 2, please click here
To read part 3, please click here
To read part 4, please click here







Deploying Models

This step is also known as inferencing or scoring a model. You can clearly see the deployment and operation of an ML pipeline when the model is tested on live data in production, which is generally done to get deeper insight and data to improve the model continuously. So, if you collect the model's performance over time, it will guarantee the model's improvement in it. There are two main architectures for ML-scoring pipelines:
  • Batch scoring using pipelines- It's an offline process in which you can evaluate an ML model against a batch of data. The result of this scoring is usually not time-critical, and the to be scored data is also larger than the model.

  • Real-time scoring using a container-based web service endpoint- It's a technique in which you can score single data inputs (common in stream processing for scoring single events in real time), which makes this task highly time-critical, and the execution is blocked until the resulting score is computed. 

Developing & Operating Enterprise-Grade ML Solutions

In order to operationalize ML projects, we have to use automated pipelines and development-operations (DevOps) methodologies like Continuous Integration (CI) and Continuous Delivery/Continuous Deployment (CD), also known as MLOps. Hence, the two different automated pipelines are:
  • Training pipeline- This one contains loading datasets, transformation, model training, and registering final models; and could be triggered via changes in the datasets or possible detected data drift in a deployed model.

  • Deployment pipeline- In this one, we have loading of models from the registry, creating and deploying Docker images, creating and deploying operational scripts, and the final deployment of the model to the target; and could be triggered by any new versions of an ML model.

Now, after having these pipelines, we can look upon Azure DevOps (and other tooling) to create a life cycle for our ML projects containing the following parts:

  • Creating or retraining a model- Here, training pipelines are used to create or retrain our model while version-controlling the pipelines and the code.

  • Deploying the model & creating scoring files & dependencies- In this one, we can use a deployment pipeline to deploy a specific mode version while version-controlling the pipeline and the code.

  • Creating an audit trail- Through CI/CD pipelines and version control, we can create an audit trail for all assets while ensuring integrity and compliance.

  • Monitoring model in production- Here, we can monitor the performance and possible data drift, which might trigger retraining of the model automatically. 








To read part 1, please click here
To read part 2, please click here
To read part 3, please click here
To read part 4, please click here
















































Comments

Post a Comment

Popular posts from this blog

Query, Visualize, & Monitor Data in Azure Sentinel

By Ashwin Venugopal -
Image
  Azure Sentinel Workbooks Several ready for use templates are provided by the Azure Sentinel that can be used to create your own workbook and then modify them as needed for Contoso. Most of the data connectors it uses to ingest data come with their own workbooks, but better insight can be obtained by simply looking into the data that is being ingested by using tables and visualizations, including bar and pie charts. You can also make your own workbook easily by using this data from the beginning instead of using the predefined templates. Workbook Page You can easily access the workbook page from the Azure Sentinel from the navigation pane which consists of the: Workbook header- You can add a new workbook and review the saved workbooks as well as templates that are available on the workbook page. Templates section- You can access existing workbook templates on the Templates tab. You can save some of the workbooks for quick access and they will appear on the My Workbooks tab.  From the
Read more

Planning for Implementing SAP Solutions on Azure (Part 2 of 5)

By Ashwin Venugopal -
Image
  To read part 1 please click  here To read part 3 please click  here To read part 4 please click  here To read part 5 please click  here Load Balancing The load balancer uses probe ports to determine as well as route the traffic to an active DBMS node, but of there's any failover then the most common SAP application architecture can automatically reconnect against the private virtual IP address without the need for the SAP application to reconfigure it, while the load balancer redirect the traffic against the private virtual IP address without any need for the SAP application to reconfigure. The loaf balancer also offers an option of DirectServerReturn which if configured, can help in directing the traffic from the DBMS VM to the SAP application layer directly to the application layer without routing through the load balancer. But, if it isn't configured, then the return traffic to the SAP application layer is routed through the load balancer. You can also use Azure Load Balan
Read more

Work with String Data Using KQL Statements

By Ashwin Venugopal -
Image
  Extract Data from Unstructured String Fields Unstructured string fields often contains the Security log data and requires parsing to extract data. There are multiple ways of pulling information from string fields in KQL and the two primary operators used are extract and parse.  Extract It gets a match for a regular expression from a text string. You can convert the extracted substring to the indicated type.  Arguments regex- A regular expression. captureGroup- A positive int constant indicating the capture group to extract. 0 stands for the entire match, 1 for the value matched by the first '('parenthesis')' in the regular expression, 2 or more for subsequent parenthesis.  text- A string to search. typeLiteral- An optional type literal. If provided, the extracted substring is converted to this type.  Returns If regex finds a match in text- the substring matched against the indicated capture group captureGroup, optionally converted to typeLiteral. If there's no mat
Read more