Build with an Azure free account. Get USD200 credit for 30 days and 12 months of free services.

Start free today

Training Machine Learning models with ML.NET

Play Training Machine Learning models with ML.NET
Sign in to queue

Description

ML.NET allows .NET developers to easily build and also consume machine learning models in their NET applications.

In this episode, Bri Achtman joins Rich to show off some really interesting scenarios that ML.NET and its family of tools enables. They talk about training models, AutoML, the ML.NET CLI, and even a Visual Studio Extension for training models!

  • [01:40] - What is ML .NET?
  • [05:19] - How can I load my data into ML .NET?
  • [06:55] - Sentiment analysis  model creation demo
  • [10:54] - Model training and consumption
  • [15:30] - Rich's ML validation test
  • [16:37] - Object detection demo
  • [18:53] - How are customers using ML .NET?
  • [22:21] - Using AutoML and the Model Builder extension for Visual Studio
  • [25:06] - Using AutoML with the ML .NET CLI

Useful Links

 

Embed

Download

The Discussion

  • User profile image
    JalisRahat
    scenarios that ML.NET and its family of tools enables. They talk about training models, AutoML, the ML.NET CLI, and even a Visual Studio Extension for training models!
  • User profile image
    adi

    can we use it using vb.net ?

  • User profile image
    Dipti

    I am using Microsoft SQL 2016 to get the required data for ML and then using binary classification to train model and doing predictions.
    I want to how to automate the process to train model biweekly on the new data from microsoft sql 2016,
    Is there any pipeline which can be used? Or I need to do this manually generate the zip file and consume at the application end?

  • User profile image
    tauheedul

    can we use it using vb.net ?



    @adi:

    Yeah it works with .NET languages... C sharp, VB.NET and F sharp
  • User profile image
    briacht
    @adi:

    Yes! You can check out some ML.NET VB samples from the community here: https://github.com/Nukepayload2/machinelearning-samples/tree/master/samples/visualbasic
  • User profile image
    cesardl
    @Dipti:

    You can directly train an ML.NET model against data in a relational database such as SQL Server.
    In the current GA version you can do that through the LoadFromEnumerable() API. you can see an example of it here: https://github.com/dotnet/machinelearning-samples/tree/master/samples/csharp/getting-started/DatabaseIntegration

    However, next week (early September) we're going to release an even simpler Relational Database Loader, so this new Database Loader provides a much simpler code implementation for you since the way it reads from the database and makes data available through the IDataView is provided out-of-the-box by the ML.NET framework so you just need to specify your database connection string, what’s the SQL statement for the dataset columns and what’s the data-class to use when loading the data. It is that simple!

    Stay tuned and check out this new feature later this week.

    And by the way, I'm interested on your SQL Server scenario about "how to automate the database process to train model biweekly". Can you send me an email to cesardl at microsoft.com so we can help you out and you provide feedback as well? :)

  • User profile image
    Dipti

    Thanks for the update, I sent you an email with the scenario

  • User profile image
    jahu
    Hi!
    @briacht
    Could you please help me to understand whether or not it is possible to train DNN from scratch without any Transfer using ML.NET? In our scenario we have hyperspectral images from plants and the DNN exsample only gives very poor classification results.
    Thynk you!
    Jan
  • User profile image
    Zeeshan Siddiqui

    Hi @jahu,

    Our current Image Classification does not support model retrain, i.e it only trains the last layer. With in the next month we plan to add the capability to also fine tune the model by allowing re-train of layers. In the meantime can I please get a link to your dataset so that I can start testing it on my side? please provide train and validation set and the anticipated accuracy, on this note have you trained this with another framework where you were getting better accuracy?

    Thanks,
    Zeeshan Siddiqui

  • User profile image
    jahu
    @Zeeshan Siddiqui:

    Hi @Zeeshan!
    Can we comunicate on a private track?

    Thanks!
    Jan

  • User profile image
    Zeeshan Siddiqui

    Hi Jan,

    Yes, does chatting on dotnet/machinelearning gitter channel(https://gitter.im/dotnet/mlnet) work? You can private message me there, my alias is codemzs on github/gitter.

    Thanks,
    Zeeshan Siddiqui

  • User profile image
    mstewartaiz
    @Zeeshan Siddiqui:

    Hello Zeeshan,

    Does ML.NET support reinforcement learning? If so, are there any examples? If not, are there any plans to support this in the future?

    Thanks,
    Michael

Add Your 2 Cents