Machine Learning and AI with ML.NET

• Knowledge of machine learning algorithms and libraries
• Strong command of C# programming language
• Experience with .NET development platforms
• Basic understanding of data science tools
• Experience with basic machine learning applications

Audience

• Data Scientists
• Machine Learning Developers

ML.NET is a framework for machine learning applications built upon the .NET development platform by Microsoft. It is extensible to multiple operating systems and can perform various machine learning prediction tasks. The central ML.NET tools are the ML.NET CLI and Model Builder which allow .NET developers to generate, train, and deploy machine learning models based on the objectives of an enterprise.

This instructor-led, live training (online or onsite) is aimed at data scientists and developers who wish to use ML.NET machine learning models to automatically derive projections from executed data analysis for enterprise applications.

By the end of this training, participants will be able to:

• Install ML.NET and integrate it into the application development environment.
• Understand the machine learning principles behind ML.NET tools and algorithms.
• Build and train machine learning models to perform predictions with the provided data smartly.
• Evaluate the performance of a machine learning model using the ML.NET metrics.
• Optimize the accuracy of the existing machine learning models based on the ML.NET framework.
• Apply the machine learning concepts of ML.NET to other data science applications.

Format of the Course

• Interactive lecture and discussion.
• Lots of exercises and practice.
• Hands-on implementation in a live-lab environment.

Course Customization Options

• To request a customized training for this course, please contact us to arrange.

Introduction

Installing and Configuring Machine Learning for .NET Development Platform (ML.NET)

• Setting up ML.NET tools and libraries
• Operating systems and hardware components supported by ML.NET

Overview of ML.NET Features and Architecture

• The ML.NET Application Programming Interface (ML.NET API)
• ML.NET machine learning algorithms and tasks
• Probabilistic programming with Infer.NET
• Deciding on the appropriate ML.NET dependencies

Overview of ML.NET Model Builder

• Integrating the Model Builder to Visual Studio
• Utilizing automated machine learning (AutoML) with Model Builder

Overview of ML.NET Command-Line Interface (CLI)

• Automated machine learning model generation
• Machine learning tasks supported by ML.NET CLI

Acquiring and Loading Data from Resources for Machine Learning

• Utilizing the ML.NET API for data processing
• Creating and defining the classes of data models
• Annotating ML.NET data models
• Cases for loading data into the ML.NET framework

Preparing and Adding Data Into the ML.NET Framework

• Filtering data models for with ML.NET filter operations
• Working with ML.NET DataOperationsCatalog and IDataView
• Normalization approaches for ML.NET data pre-processing
• Data conversion in ML.NET
• Working with categorical data for ML.NET model generation

Implementing ML.NET Machine Learning Algorithms and Tasks

• Binary and Multi-class ML.NET classifications
• Regression in ML.NET
• Grouping data instances with Clustering in ML.NET
• Anomaly Detection machine learning task
• Ranking, Recommendation, and Forecasting in ML.NET
• Choosing the appropriate ML.NET algorithm for a data set and functions
• Data transformation in ML.NET
• Algorithms for improved accuracy of ML.NET models

Training Machine Learning Models in ML.NET

• Building an ML.NET model
• ML.NET methods for training a machine learning model
• Splitting data sets for ML.NET training and testing
• Working with different data attributes and cases in ML.NET
• Caching data sets for ML.NET model training

Evaluating Machine Learning Models in ML.NET

• Extracting parameters for model retraining or inspecting
• Collecting and recording ML.NET model metrics
• Analyzing the performance of a machine learning model

Inspecting Intermediate Data During ML.NET Model Training Steps

Utilizing Permutation Feature Importance (PFI) for Model Predictions Interpretation

Saving and Loading Trained ML.NET Models

• ITTransformer and DataViewScheme in ML.NET
• Loading locally and remotely stored data
• Working with machine learning model pipelines in ML.NET

Utilizing a Trained ML.NET Model for Data Analyses and Predictions

• Setting up the data pipeline for model predictions
• Single and Multiple predictions in ML.NET

Optimizing and Re-training an ML.NET Machine Learning Model

• Re-trainable ML.NET algorithms
• Loading, extracting and re-training a model
• Comparing re-trained model parameters with previous ML.NET model

Integrating ML.NET Models with The Cloud

• Deploying an ML.NET model with Azure functions and web API

Troubleshooting

Summary and Conclusion