{"id":24686,"date":"2019-09-24T12:48:06","date_gmt":"2019-09-24T19:48:06","guid":{"rendered":"https:\/\/devblogs.microsoft.com\/dotnet\/?p=24686"},"modified":"2019-09-24T13:03:48","modified_gmt":"2019-09-24T20:03:48","slug":"ml-net-and-model-builder-at-net-conf-2019-machine-learning-for-net","status":"publish","type":"post","link":"https:\/\/devblogs.microsoft.com\/dotnet\/ml-net-and-model-builder-at-net-conf-2019-machine-learning-for-net\/","title":{"rendered":"ML.NET and Model Builder at .NET Conf 2019 (Machine Learning for .NET)"},"content":{"rendered":"

We are excited today to announce updates to Model Builder<\/a> and improvements in ML.NET<\/a>. You can learn more in the \u201cWhat\u2019s new in ML.NET<\/a>?.\u201d session at .NET Conf<\/a>.<\/p>\n

ML.NET<\/a> is an open-source and cross-platform machine learning framework (Windows, Linux, macOS) for .NET developers.<\/p>\n

ML.NET<\/a> offers Model Builder Model Builder<\/a> (a simple UI tool) and CLI<\/a> to make it super easy to build custom ML Models using AutoML.<\/p>\n

Using ML.NET<\/a>, developers can leverage their existing tools and skillsets to develop and infuse custom AI into their applications by creating custom machine learning models for common scenarios like Sentiment Analysis, Recommendation, Image Classification and more!.<\/p>\n

Following are the key highlights:<\/p>\n

Model Builder updates<\/h2>\n

This release of Model Builder adds support for a new scenario and address many customer reported issues.<\/p>\n

\"Model<\/p>\n

Feature engineering<\/strong>: In previous versions of Model Builder, after selecting your dataset, either from a file or from SQL Server, you only had the option to choose the column to predict (the Label). Any other columns in the dataset were automatically used to make the prediction (Features). Any columns that you did not want to include, you had to manipulate your dataset outside of Model Builder and then upload the modified dataset.<\/p>\n

\"Feature<\/p>\n

Model consumption made easy!<\/strong>: In previous versions of Model Builder, there were numerous steps that you had to take after Model Builder\u2019s code and model generation in order to consume the trained model in your app, including adding a reference to the generated library project, setting the model Copy to Output property to \u201cCopy If Newer,\u201d and adding the Microsoft.ML NuGet package to your app.<\/p>\n

This has all been simplified and automated, so now all you have to do is copy + paste the code from the Next Steps in Model Builder, and then you can run your app and start making predictions!<\/p>\n

Address customer feedback<\/strong>: This release also address many customer reported issues around installation errors, usability feedback and stability improvements and more. Learn more here<\/a>.<\/p>\n

ML.NET<\/a> updates<\/h2>\n

This is a short summary of the features and enhancements added to ML.NET over the last few months.<\/p>\n