Microsoft has unveiled a preview of the ML.NET Textual content Classification API, an API meant to make it simpler to coach customized textual content classification fashions utilizing the open supply ML.NET machine studying framework.
Launched June 14, the ML.NET Textual content Classification API makes use of “state-of-the-art” deep studying strategies, Microsoft mentioned. ML.NET permits builders to combine customized machine studying fashions into .NET apps. Textual content classification is the method of making use of labels or classes to textual content. Frequent use instances embody categorizing e-mail as spam or not spam, analyzing sentiment as optimistic or unfavorable from buyer critiques, and making use of labels to assist tickets.
The ML.NET Textual content Classification API is powered by the TorchSharp .NET library, which supplies entry to the libtorch library that powers the PyTorch machine studying framework. TorchSharp has low-level capabilities for coaching neural networks from scratch in .NET. For ML.NET, among the complexity of TorchSharp has been abstracted to make this coaching simpler.
In collaboration with Microsoft Analysis, Microsoft took the TorchSharp implementation of NAS-BERT (Bidirectional Encoder Representations from Transformers), a variant of BERT obtained with neural structure search, and added it to ML.NET. Beginning with a pre-trained model of this mannequin, the Textual content Classification API makes use of the consumer’s information to fine-tune the present mannequin quite than to construct a brand new mannequin from scratch.
The Textual content Classification API is a part of the two.0.0 and 0.20.0 preview variations of ML.NET. Along with the Microsoft.ML bundle, it requires Microsoft.ML.TorchSharp and both TorchSharp-cpu (if utilizing a CPU) or TorchSharp-cuda-windows or TorchSharp-cuda-linux (if utilizing a GPU).
Builders can use the NuGet bundle supervisor in Visible Studio or the .NET CLI to put in the packages. Code samples of the API will be discovered within the Textual content Classification API Pocket book.
Microsoft pointed on the market are nonetheless limitations with the API corresponding to not with the ability to use the Consider methodology to calculate analysis metrics. Enhancements are deliberate to the API together with introducing different scenario-based APIs.
Copyright © 2022 IDG Communications, Inc.
