We are pleased to announce the release of updates to Model Builder and ML.NET ML.NET – is a cross-platform open-source machine learning environment (Windows, Linux, macOS) for .NET developers.
ML.NET. offers Model Builder (a simple user interface tool) and command line interface , designed to make it easy to create custom ML models using AutoML.
Using ML.NET, developers can leverage their existing tools and skill sets to develop and implement AI into applications, creating custom machine learning models for common scenarios such as text tone analysis, recommendations, image classification, and more!
Next up are the major novelties :
Model Builder updates
This release of Model Builder adds new script support and fixes many issues reported by users.
Features Development : In previous versions of Model Builder, after selecting a dataset from a file or from SQL Server, you could only select a column for prediction (Label).Any other columns in the dataset were automatically used for prediction (Features).Any columns you didn’t want to include had to be changed in the dataset itself outside of Model Builder, and then load the changed dataset.
In previous versions of Model Builder, there were many steps you had to take after generating code and Model Builder to use the trained model in your application, including adding a reference to the library project you created, setting the "Copy to Output" property to "Copy If Newer" and adding the Microsoft.ML NuGet package to your application.
This has all been simplified and automated, so now all you have to do is copy + paste the code from Next Steps into Model Builder and then you can run your application.
This is a summary of features and enhancements added to ML.NET over the past few months.
- Support for .NET Core 3
- Support for new scenarios such as sales forecasting, anomaly detection
- Preview : a proprietary database loader that allows for training directly on relational databases
- Preview : Creating custom deep learning models for image classification using TensorFlow.
We’ve worked hard to add more documentation and tutorials, guides, and more for the Model Builder, CLI, and ML.NET Framework.We’ve also simplified the table of contents for documents ML.NET , so you can easily find the data you need.
New training modules for ML.NET
To help users learn the basics of machine learning and ML.NET, we’ve created a set of tutorial videos.You can watch them here
A wide range of examples to study
We have added many scenarios for different use cases with machine learning. You can explore and customize these samples for your scenario. You can find more samples in the repository ML.NET Samples on GitHub