to evaluate The mood in the stock market or the price forecasts for customers? With ML.NET 1.4 can .NET-Developers and developers using the model Generator in Visual Studio automatically generated a series of scenarios for machine Learning as a model, even without ML knowledge.
The cross-platform Machine Learning Framework ML.NET appeared in the Version 1.4. The latest Release contains new Features of this extension .NET Core 3.0, it supports in the future Jupyter Notebooks and native DNN Retraining in the case of the image classification with GPU.
what's new for ML.NET 1.4
ML.NET 1.4 can now optionally also in a Build .NET Core 3.0 will be used. This Feature was included in the previous Preview, but is now for the first time, widely available. ML.NET can still older in .NET versions are used. Details on the Developer Blogs from Microsoft.</stand p>
The images classification is now based on DNN Retraining with GPU support. This Feature allows the DNN to the transfer of learning ML.NET on the Basis of the image data sets and Image maps. With this Feature, you can generate an image classification model. The low-level procedures Tensorflow.NET Library in the past required several hundred lines of Code. ML.NET 1.4 offers the advantage of a high-level API, which you define with a few lines of C#Code just as image classification model can teach. The APIs of some of the vortrainierter tensor flow models (DNN-architectures) are available for .NET is now more accessible.
The GPU support for Windows and Linux is new. More recently, an earlier Finish of the workout is optional, once the learning goal is reached. The schedule for the Learning rate can now be controlled with more precision – more to the event planner for the Learning on the pages of tensor flow.</guess p>
ML.NET 1.4 is also supported by Jupyter Notebooks – the new .NET Kernel for Jupyter allows this. Jupyter Notebooks can any .NET Code to process (C#, F#), and therefore ML.NET-Code. Jupyter Notebook is an Open-Source web application, with the help of which you can share texts, which contain Live Code, visualizations and explanatory Text. More on this on the pages of Jupyter.
The core components of ML.NET 1.4 support more different processor architectures. On modern processors ml.net can train simply faster because it can execute several Floating-Point processes at the same time, as in the previous C++ Code, the only supported SSE instructions. Hardware-intrinsic C#Code brings advantages: ML.NET can run as a Fallback-mode mathematical processes, number for number, if the processor supports neither SSE nor AVX (what Chips is, for example, in the case of ARM the case).
new features in ML.NET model-Generator
The preview version of the ML.NET model generator contains two new Features: For the image classification model with your own pictures can be trained locally (image classification scenario). This can be done directly via the web interface. Customer-specific scenarios could say, for example, the detection of fraudulent Bank transactions, price forecasts, or the forwarding of customer feedback to the appropriate Team. With automated machine Learning (AutoML) can .NET Developer and developer in Visual Studio without ML knowledge of simple models to generate.
Additional Links, tips, and Tutorials
An illustrated guide in steps to ML.NET the model Generator can be found on the pages of Telerik. A Tutorial for ML.NET is in the .NET-the Blog of Microsoft. Currently, the model Generator is still in the preview phase. The ML.NET the model Generator uses automated machine Learning (AutoML). ML.NET with the model-Generator can be downloaded on the pages of Visual Studio for free. There is also a step-by-step instructions. On the DevBlogs tips to Improve the performance of the selected model and targeted Training with Algorithms. The larger the data base, fall the more accurate the forecasts.
How to speed up machine Learning, by ML.NET the new Hardware-intrinsic APIs .NET Core 3.0 uses, Brian Lui explains in his Blog. The list of all new features and improvements is on the DevBlogs from Microsoft for future Reference.