book

Mastering Azure Machine Learning

Name: Mastering Azure Machine Learning
ISBN: 9781789807554

by Christoph Körner, Kaijisse Waaijer

April 2020

Intermediate to advanced

436 pages

10h 16m

English

Packt Publishing

Read now

Unlock full access

Preface
About Mastering Azure Machine LearningAbout the authorsAbout the reviewersLearning objectivesAudienceApproachTo get the most out of this bookConventionsDownload resources
Section 1: Azure Machine Learning
1. Building an end-to-end machine learning pipeline in Azure
Performing descriptive data explorationMoving data to the cloudUnderstanding missing valuesVisualizing data distributionsFinding correlated dimensionsMeasuring feature and target dependencies for regressionVisualizing feature and label dependency for classificationExploring common techniques for data preparationLabeling the training dataNormalization and transformation in machine learningEncoding categorical variablesA feature engineering example using time-series dataUsing NLP to extract complex features from textChoosing the right ML model to train dataChoosing an error metricThe training and testing splitAchieving great performance using tree-based ensemble modelsModeling large and complex data using deep learning techniquesOptimization techniquesHyperparameter optimizationModel stackingAzure Automated Machine LearningDeploying and operating modelsBatch scoring using pipelinesReal-time scoring using a container-based web serviceTracking model performance, telemetry, and data skewSummary
2. Choosing a machine learning service in Azure
Demystifying the different Azure services for MLChoosing an Azure service for MLChoosing a compute target for Azure Machine LearningAzure Cognitive Services and Custom VisionAzure Cognitive ServicesCustom Vision—customizing the Cognitive Services APIAzure Machine Learning with GUIsAzure Machine Learning designerAzure Automated Machine LearningMicrosoft Power BIAzure Machine Learning workspaceOrganizing experiments and models in Azure Machine LearningDeployments through Azure Machine LearningSummary
Section 2: Experimentation and Data Preparation
3. Data experimentation and visualization using Azure
Preparing your Azure Machine Learning workspaceSetting up the ML Service workspaceRunning a simple experiment with Azure Machine LearningLogging metrics and tracking resultsScheduling and running scriptsAdding cloud compute to the workspaceVisualizing high-dimensional dataTracking figures in experiments in Azure Machine LearningUnsupervised dimensionality reduction with PCAUsing LDA for supervised projectionsNon-linear dimension reduction with t-SNEGeneralizing t-SNE with UMAPSummary
4. ETL, data preparation, and feature extraction
Managing data and datasets in the cloudGetting data into the cloudManaging data in Azure Machine LearningExploring data registered in Azure Machine LearningPreprocessing and feature engineering with Azure Machine Learning DataPrepParsing different data formatsBuilding a data transformation pipeline in Azure Machine LearningSummary
5. Azure Machine Learning pipelines
Benefits of pipelines for ML workflowsWhy build pipelines?What are Azure Machine Learning pipelines?Building and publishing an ML pipelineCreating a simple pipelineConnecting data inputs and outputs between stepsPublishing, triggering, and scheduling a pipelineParallelizing steps to speed up large pipelinesReusing pipeline steps through modularizationIntegrating pipelines with other Azure servicesBuilding pipelines with the Azure Machine Learning designerAzure Machine Learning pipelines in Azure Data FactoryAzure Pipelines for CI/CDSummary
6. Advanced feature extraction with NLP
Understanding categorical dataComparing textual, categorical, and ordinal dataTransforming categories into numeric valuesCategories versus textBuilding a simple bag-of-words modelA naive bag-of-words model using countingTokenization – turning a string into a list of wordsStemming – rule-based removal of affixesLemmatization – dictionary-based word normalizationA bag-of-words model in scikit-learnLeveraging term importance and semanticsGeneralizing words using n-grams and skip- gramsReducing word dictionary size using SVDMeasuring the importance of words using tf-idfExtracting semantics using word embeddingsImplementing end-to-end language modelsEnd-to-end learning of token sequencesState-of-the-art sequence-to-sequence modelsText analytics using Azure Cognitive ServicesSummary
Section 3: Training Machine Learning Models

7. Building ML models using Azure Machine Learning
Working with tree-based ensemble classifiersUnderstanding a simple decision treeCombining classifiers with baggingOptimizing classifiers with boosting roundsTraining an ensemble classifier model using LightGBMLightGBM in a nutshellPreparing the dataSetting up the compute cluster and execution environmentBuilding a LightGBM classifierScheduling the training script on the Azure Machine Learning clusterSummary
8. Training deep neural networks on Azure
Introduction to deep learningWhy DL?From neural networks to DLComparing classical ML and DLTraining a CNN for image classificationTraining a CNN from scratch in your notebookGenerating more input data using augmentationMoving training to a GPU cluster using Azure Machine Learning computeImproving your performance through transfer learningSummary
9. Hyperparameter tuning and Automated Machine Learning
Hyperparameter tuning to find the optimal parametersSampling all possible parameter combinations using grid searchTrying random combinations using random searchConverging faster using early terminationOptimizing parameter choices using Bayesian optimizationFinding the optimal model with Azure Automated Machine LearningAdvantages and benefits of Azure Automated Machine LearningA classification exampleSummary
10. Distributed machine learning on Azure
Exploring methods for distributed MLTraining independent models on small data in parallelTraining a model ensemble on large datasets in parallelFundamental building blocks for distributed MLSpeeding up DL with data-parallel trainingTraining large models with model-parallel trainingUsing distributed ML in AzureHorovod—a distributed DL training frameworkImplementing the HorovodRunner API for a Spark jobRunning Horovod on Azure Machine Learning computeSummary
11. Building a recommendation engine in Azure
Introduction to recommender enginesContent-based recommendationsMeasuring similarity between itemsFeature engineering for content-based recommendersContent-based recommendations using gradient boosted treesCollaborative filtering—a rating-based recommendation engineWhat is a rating? Explicit feedback as opposed to implicit feedbackPredicting the missing ratings to make a recommendationScalable recommendations using ALS factorizationCombining content and ratings in hybrid recommendation enginesBuilding a state-of-the-art recommender using the Matchbox RecommenderAutomatic optimization through reinforcement learningAn example using Azure Personalizer in PythonSummary
Section 4: Optimization and Deployment of Machine Learning Models
12. Deploying and operating machine learning models
Deploying ML models in AzureUnderstanding the components of an ML modelRegistering your models in a model registryCustomizing your deployment environmentChoosing a deployment target in AzureBuilding a real-time scoring serviceImplementing a batch scoring pipelineInference optimizations and alternative deployment targetsProfiling models for optimal resource configurationPortable scoring through the ONNX runtimeFast inference using FPGAs in AzureAlternative deployment targetsMonitoring Azure Machine Learning deploymentsCollecting logs and infrastructure metricsTracking telemetry and application metricsSummary
13. MLOps—DevOps for machine learning
Ensuring reproducible builds and deploymentsVersion-controlling your codeRegistering snapshots of your dataTracking your model metadata and artifactsScripting your environments and deploymentsValidating your code, data, and modelsRethinking unit testing for data qualityIntegration testing for MLEnd-to-end testing using Azure Machine LearningContinuous profiling of your modelSummary
14. What's next?
Understanding the importance of dataThe future of ML is automatedChange is the only constant – preparing for changeFocusing first on infrastructure and monitoringControlled rollouts and A/B testingSummary
Index

Content preview from Mastering Azure Machine Learning

9. Hyperparameter tuning and Automated Machine Learning

In the previous chapter, we learned how to train convolutional and more complex deep neural networks (DNNs). When training these models, we are often confronted with complex choices when parametrizing them, involving various parameters such as the number of layers, the order of layers, regularization, batch size, learning rate, the number of epochs, and more. This is not only true for DNNs; the same problem arises with selecting the correct preprocessing steps, features, models, and parameters in statistical ML approaches.

In this chapter, we will take a look at optimizing the training process in order to take away some of those error-prone human choices from machine learning. These ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Machine Learning In The Cloud With Azure Machine Learning

Publisher Resources

ISBN: 9781789807554

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Mastering Azure Machine Learning

by Christoph Körner, Kaijisse Waaijer

9. Hyperparameter tuning and Automated Machine Learning

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.