LearningModel Class

Definition

Namespace:

Windows.AI.MachineLearning

Represents a trained machine learning model.

public ref class LearningModel sealed : IClosable

/// [Windows.Foundation.Metadata.ContractVersion(Windows.AI.MachineLearning.MachineLearningContract, 65536)]
/// [Windows.Foundation.Metadata.MarshalingBehavior(Windows.Foundation.Metadata.MarshalingType.Agile)]
/// [Windows.Foundation.Metadata.Threading(Windows.Foundation.Metadata.ThreadingModel.Both)]
class LearningModel final : IClosable

[Windows.Foundation.Metadata.ContractVersion(typeof(Windows.AI.MachineLearning.MachineLearningContract), 65536)]
[Windows.Foundation.Metadata.MarshalingBehavior(Windows.Foundation.Metadata.MarshalingType.Agile)]
[Windows.Foundation.Metadata.Threading(Windows.Foundation.Metadata.ThreadingModel.Both)]
public sealed class LearningModel : System.IDisposable

Public NotInheritable Class LearningModel
Implements IDisposable

Inheritance

Object Platform::Object IInspectable LearningModel

Attributes

ContractVersionAttribute MarshalingBehaviorAttribute ThreadingAttribute

Implements

IClosable IDisposable

Windows requirements

Device family	Windows 10, version 1809 (introduced in 10.0.17763.0)
API contract	Windows.AI.MachineLearning.MachineLearningContract (introduced in v1.0)

Examples

The following example loads a model and creates an evaluation session with it.

private async Task LoadModelAsync(string _modelFileName)
{
    LearningModel _model;
    LearningModelSession _session;

    try
    {
        // Load and create the model
        var modelFile = 
            await StorageFile.GetFileFromApplicationUriAsync(new Uri($"ms-appx:///Assets/{_modelFileName}"));
        _model = await LearningModel.LoadFromStorageFileAsync(modelFile);

        // Create the evaluation session with the model
        _session = new LearningModelSession(_model);

    }
    catch (Exception ex)
    {
        StatusBlock.Text = $"error: {ex.Message}";
        _model = null;
    }
}

Remarks

This is the main object you use to interact with Windows ML. You use it to load, bind, and evaluate trained ONNX models:

1. Load the model using one of the Load* constructors.
2. Enumerate the InputFeatures and OutputFeatures and bind to your model.
3. Create a LearningModelSession and evalaute.

Windows Server

To use this API on Windows Server, you must use Windows Server 2019 with Desktop Experience.

Thread safety

This API is thread-safe.

Properties

Author	The name of the model author.
Description	A description of the model.
Domain	The domain of the model.
InputFeatures	A list of the model's input features.
Metadata	The raw <string,string> metadata from the ONNX model.
Name	The name of the model.
OutputFeatures	A list of the model's output features.
Version	The version of the model.

Methods

Close()	Releases the in-memory ONNX model. After calling this, you cannot create any more LearningModelSession objects.
Dispose()	Performs application-defined tasks associated with freeing, releasing, or resetting unmanaged resources.
LoadFromFilePath(String)	Loads an ONNX model from a file on disk.
LoadFromFilePath(String, ILearningModelOperatorProvider)	Loads an ONNX model from a file on disk.
LoadFromStorageFileAsync(IStorageFile)	Loads an ONNX model from an IStorageFile asynchronously.
LoadFromStorageFileAsync(IStorageFile, ILearningModelOperatorProvider)	Loads an ONNX model from an IStorageFile asynchronously.
LoadFromStream(IRandomAccessStreamReference)	Loads an ONNX model from a stream.
LoadFromStream(IRandomAccessStreamReference, ILearningModelOperatorProvider)	Loads an ONNX model from a stream.
LoadFromStreamAsync(IRandomAccessStreamReference)	Loads an ONNX model from a stream asynchronously.
LoadFromStreamAsync(IRandomAccessStreamReference, ILearningModelOperatorProvider)	Loads an ONNX model from a stream asynchronously.

See also

• Windows ML
• Windows ML samples (GitHub)