Shulou(Shulou.com)12/24 Report--

CTOnews.com, November 29 (Xinhua)-- Microsoft has announced the launch of ML.NET 3.0, a cross-platform machine learning framework, which mainly strengthens deep learning, improves ML.NET data processing capabilities, and adds Intel oneDAL accelerated training technology and automatic machine learning.

Microsoft CTOnews.com, the ▲ image source, notes that ML.NET 3.0 provides a number of deep learning features, including "object detection", "named entity recognition" and "question and answer processing".

Among them, "object detection" can locate and classify different types of entities in the image. According to officials, object detection is a computer vision task, which is closely related to "image classification", but the classification is relatively finer. When the image contains different types of objects, officials recommend the use of relevant functions.

The named entity identification and question-and-answer processing is based on Microsoft's newly added TorchSharp API, which is a. Net library that claims to combine the latest technology of Microsoft Research and the Transformer neural network architecture in TorchSharp, and uses the existing TorchSharp RoBERTa text classification function as the basis to achieve the above functions.

In addition, shortly after the previous release of ML.NET 2.0, Microsoft announced that it would support Intel's oneDAL accelerated training technology, which is now available in ML.NET 3.0 and can significantly accelerate data analysis and machine learning.

Microsoft also updated the automatic machine learning (AutoML) function of ML.NET 3.0. it brings the functions of sentence similarity, question and answer processing and object detection, which can help developers choose the most appropriate model and parameters, and make it easier for developers to design machine learning models.

CTOnews.com also found that ML.NET now has continuous resource monitoring capability, and can monitor the usage of RAM and hard disk space through AutoML.IMonitor, making it easy for developers to control long-running experiments and avoid crashes of running processes due to insufficient RAM or ROM. At the same time, it is easy for developers to directly view the parameters of the process.

ML.NET 3.0 also integrates Tensor Primitives, a new set of API specifically for tensor computing, which can further promote the application of .NET in artificial intelligence mathematical operations. The API not only uses the internal instruction set of hardware to speed up the operation efficiency, but also combines the principle concept of generic mathematics (Generic Math). It is known as "a powerful tool for developers to deal with complex mathematics and tedious data".

▲ image source Microsoft

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.