Windows ML

Windows Machine Learning (WinML)

Overview

• Overview of Windows Machine Learning

Concept

• What is a machine learning model?
• Working with ONNX models
• Windows ML performance and memory
• Executing multiple ML models in a chain

Tutorial

• Image classification with Custom Vision and Windows Machine Learning
• Image Classification with ML.NET and Windows Machine Learning
• Image classification with PyTorch and Windows Machine Learning
• Data analysis with PyTorch and Windows Machine Learning
• Object detection with TensorFlow and Windows Machine Learning
• Create a basic WinML UWP app (C#)
• Create a basic WinML UWP app (C++)
• Convert trained models to ONNX

Reference

• API Reference
• Core APIs (Windows.AI.MachineLearning Namespace)
• Custom operators
• WinML native APIs
• Tools and Samples
• Release Notes
• Frequently Asked Questions

Direct Machine-Learning (DirectML)

Overview

• Overview of DirectML
• GPU accelerated ML training in WSL

Concept

• Binding in DirectML
• UAV barriers and resource state barriers
• Using strides to express padding, memory layout
• Resource lifetime and synchronization
• Using the debug layer
• Handling errors and device-removal
• Helper functions

Reference

• DirectML reference
• Sample applications