A Pragmatic Deep Dive into ML.NET
summary
ML.NET is presented as a local .NET machine-learning stack that trains and exports compact portable ZIP models for deployment in C# apps and mobile without cloud dependencies. The talk demonstrates concrete workflows: k-means clustering (5 clusters) for customer segmentation, a language detector trained on ~5,000 sentences, and AutoML selecting top algorithms on a Portuguese wine dataset, using Visual Studio and CLI tooling to build pipelines, evaluate with standard metrics, and export on-device models.
AI-written summary · about
chapters (39)
- 0:00Introduction
- 1:13ML.NET status
- 3:21ML.NET use cases
- 5:09ML.NET in .NET apps
- 6:32Microsoft ML.NET growth
- 8:02Barriers to ML starting
- 9:04ML.NET prerequisites
- 10:13Data prep for ML
- 11:39Training ML models
- 12:21Supervised learning
- 13:51Customer segmentation
- 16:23Clustering example
- 17:06Data prep for clustering
- 18:18Training clustering model
- 20:48Excel clustering
- 23:19Evaluating clustering
- 25:06Clustering initialization
- 26:11Supervised learning labeled
- 27:18Text classification
- 28:49Language detection
- 30:14Choosing trainer
- 31:39Finance model decisions
- 32:42Text to features
- 33:49Model quality metrics
- 35:18Alternative metrics
- 36:22Deploying ML.NET models
- 37:31Region detector
- 38:35ML.NET mobile apps
- 39:20ML.NET AutoML
- 40:50AutoML algorithm selection
- 41:56Model performance metrics
- 43:01AutoML tools
- 44:25Running AutoML experiment
- 45:29Understanding AutoML results
- 46:33Selecting best model
- 47:15Tuning ML.NET models
- 48:06Integrating ML.NET
- 48:48Recommendation systems
- 49:08Q&A
related talks
57mMachine Learning for .NET Developers with ML.NET
01h 04mIntroduction to ML.NET
45mML: ML.NET: deep dive into dev AI
01h 03mMachine learning for C# developers: Introducing ML
52mML.NET - Machine Learning made for .NET
59mMachine Learning in .NET (ML.NET)
57mMachine Learning Made Easy – An Introduction to ML.NET
58mExploring Machine Learning for .Net Developers