dotnet/machinelearning-samples

What it solves

This repository provides a comprehensive collection of practical examples to help .NET developers integrate machine learning into their applications using the ML.NET framework. It bridges the gap between the ML.NET API and real-world implementation for various common ML tasks.

How it works

The project is organized into two primary types of samples:

• Getting Started: Simple console applications focusing on specific ML tasks to demonstrate core code implementation.
• End-to-End Apps: Full web and desktop applications that show how to deploy ML models in production-like environments (e.g., WebAPI, Razor, Azure Functions, and Blazor).

It also includes preview features for automated model generation via a Command Line Interface (CLI) and an AutoML API, which automatically select algorithms and hyperparameters based on provided datasets.

Who it’s for

.NET developers who want to implement machine learning capabilities—such as classification, regression, and computer vision—within the .NET ecosystem using C# or F#.

Highlights

• Diverse ML Task Coverage: Includes samples for binary and multi-class classification, recommendation systems, regression, time series forecasting, anomaly detection, clustering, and ranking.
• Computer Vision: Examples for image classification (via high-level APIs, TensorFlow, and TensorFlow Featurizer) and object detection (via ONNX).
• AutoML Integration: Tools to automate model training and code generation for binary classification, multi-class classification, and regression.
• Deployment Examples: Demonstrates how to scale models on WebAPI, Azure Functions, and Blazor apps.