The expert guide to creating production machine learning solutions with ML.NET!
ML.NET brings the power of machine learning to all .NET developers- and Programming ML.NET helps you apply it in real production solutions. Modeled on Dino Esposito's best-selling Programming ASP.NET, this book takes the same scenario-based approach Microsoft's team used to build ML.NET itself. After a foundational overview of ML.NET's libraries, the authors illuminate mini-frameworks ("ML Tasks") for regression, classification, ranking, anomaly detection, and more. For each ML Task, they offer insights for overcoming common real-world challenges. Finally, going far beyond shallow learning, the authors thoroughly introduce ML.NET neural networking. They present a complete example application demonstrating advanced Microsoft Azure cognitive services and a handmade custom Keras network- showing how to leverage popular Python tools within .NET.
14-time Microsoft MVP Dino Esposito and son Francesco Esposito show how to:
- Build smarter machine learning solutions that are closer to your user's needs
- See how ML.NET instantiates the classic ML pipeline, and simplifies common scenarios such as sentiment analysis, fraud detection, and price prediction
- Implement data processing and training, and "productionize" machine learning-based software solutions
- Move from basic prediction to more complex tasks, including categorization, anomaly detection, recommendations, and image classification
- Perform both binary and multiclass classification
- Use clustering and unsupervised learning to organize data into homogeneous groups
- Spot outliers to detect suspicious behavior, fraud, failing equipment, or other issues
- Make the most of ML.NET's powerful, flexible forecasting capabilities
- Implement the related functions of ranking, recommendation, and collaborative filtering
- Quickly build image classification solutions with ML.NET transfer learning
- Move to deep learning when standard algorithms and shallow learning aren't enough
- "Buy" neural networking via the Azure Cognitive Services API, or explore building your own with Keras and TensorFlow
