Microsoft Releases ML.NET 3.0

Microsoft has announced the release of ML.NET 3.0, its open-source, cross-platform, and .NET-based machine learning (ML) framework. Aimed at developers who wish to build custom machine learning models and integrate them into their apps, ML.NET arrives about two weeks after the firm issued .NET 8, the latest long-term support version of its open source, cross-platform, managed software development platform.

“ML.NET version 3.0 is now released, with lots of new features and enhancements,” Microsoft’s Jeff Handley writes in the announcement post. “ML.NET is an open-source, cross-platform machine learning framework for .NET developers that enables integration of custom machine learning models into .NET applications.”

You can find the full list of improvements in the release notes, but here are some key highlights:

Deep learning improvements. ML.NET 3.0 supports new deep learning capabilities like object detection (computer vision) and named entity recognition and question answering (both of which are based on natural language processing, or NLP).

Accelerated ML.NET training. Using the Intel oneAPI Data Analytics Library (oneDAL), ML.NET 3.0 uses hardware acceleration via the SIMD extensions in 64-bit Intel and AMD microprocessors to speed up data analysis dramatically.

Automated machine learning (AutoML). This capability automates the process of applying machine learning to data, and in ML.NET 3.0 it supports Sentence Similarity, Question Answering, and Object Detection, plus continuous resource monitoring.

Looking ahead, Handley notes that the team is now working on .NET 9 and ML.NET 4.0, but that customers can expect Model Builder and ML.NET CLI updates well before then. There will also be interim updates to DataFrame and more Tensor APIs.

Developers can learn more about ML.NET at Microsoft Learn.