Last week, the team behind ML.NET announced the release of ML.NET 1.0 RC (Release Candidate), an open-source and cross-platform machine learning framework for .NET developers. ML.NET 1.0 RC is the last preview release before releasing the final ML.NET 1.0 RTM (Requirements Traceability Matrix) this year. Developers can use ML.NET in sentiment analysis, product recommendation, spam detection, image classification, and much more.
What’s new in ML.NET 1.0 RC?
Preview packages
According to the Microsoft blog, “Heading ML.NET 1.0, most of the functionality in ML.NET (around 95%) is going to be released as stable (version 1.0).”
The packages that will be available for the preview state are TensorFlow, Onnx components, TimeSeries components, and recommendation components.
IDataView moved to Microsoft.ML namespace
In this release, IDataView has been moved back into Microsoft. ML namespace based on feedback the team received.
Support for TensorFlow models
This release comes with added support for TensorFlow models, an open source machine learning framework used for deep learning projects. The issues in ML.NET version 0.11 related to TensorFlow models have been fixed in this release.
Major changes in ML.NET 1.0 RC
The ‘Data’ namespace has been removed in this release with the help using Microsoft.Data.DataView. The Nuget package has been added for Microsoft.ML.FastTree. Also, PoissonRegression has been changed to LbfgsPoissonRegression.
To know more about this release, check out the official announcement.
Read Next
.NET team announces ML.NET 0.6
Qml.Net: A new C# library for cross-platform .NET GUI development
ML.NET 0.4 is here with support for SymSGD, F#, and word embeddings transform!A