Cloud Native AI and Machine Learning on AWS

Preface

Introducing the ML Workflow

• Introduction
• Evolution of AI and ML
• Approaching an ML problem
• Overview of the ML workflow
• Introducing AI and ML on AWS
• Navigating the ML workflow
• Conclusion
• Points to Remember

Hydrating the Data Lake

• Introduction
• Lesson Scenario
• The Data Lake
• Securing your Buckets
• Securing your Data Lake
• Data Lakes for Machine Learning
• The Importance of Hydration
• Setting Up Your AWS Account
• Starting Datasets
• Streaming Data and the Data Lake
• Uncovering Patterns
• Amazon Athena
• Conclusion
• Points to Remember

Predicting the Future With Features

• Introduction
• Technical Requirements
• Introducing feature engineering
• Tokenize and remove punctuations
• Feature engineering for computer vision
• Resizing Images
• Cropping and tiling images
• Rotating images
• Converting to grayscale
• Converting to RecordIO format
• Dimensionality reduction with Principal Component Analysis
• Feature engineering for tabular datasets
• Exploring the data
• Imputing missing values
• Feature selection
• Feature frequency encoding
• Target mean encoding
• One hot encoding
• Feature scaling
• Feature normalization
• Binning
• Feature correlation
• Principal Component Analysis
• Conclusion
• Points to Remember

Orchestrating the Data Continuum

• Introduction
• Demystifying the data continuum
• Running feature engineering with AWS Glue ETL
• Data profiling with AWS Glue DataBrew
• Conclusion
• Points to Remember

Casting a Deeper Net (Algorithms and Neural Networks)

• Introduction
• Introducing Algorithms and Neural networks
• Simplifying the Algorithm versus Neural network conundrum
• Building ML solutions with Algorithms and Neural Networks
• Conclusion
• Points to Remember

Iteration Makes Intelligence (Model Training and Tuning)

• Introduction
• The Meaning of Training
• What Training Means for Deep Learning
• GPU vs CPU
• AWS Trainium
• Transfer Learning
• The Mise en Place of Model Training
• Defining Model Training and Evaluation Metrics
• Setting Up Model Hyperparameters
• Script vs Container
• Training Data Storage and Compute
• Training Scenarios
• Linear Regression
• Natural Language Processing
• Image Classification
• Conclusion
• Points to Remember

Let George Take Over (AutoML in Action)

• Introduction
• Running AutoML with SageMaker Canvas
• Automated Hyperparameter Tuning
• Using AutoGluon for AutoML
• Conclusion
• Points to Remember

Blue or Green (Model Deployment Strategies)

• Introduction
• Inference Options
• Choosing your Compute
• Amazon SageMaker Endpoint
• Inference at the Edge
• Deployment Mechanics
• After the Deployment
• Updating a Deployed Model
• Conclusion
• Points to Remember

Wisdom at Scale with Elastic Inference

• Introduction
• Understanding SageMaker ML Inference options
• SageMaker endpoints for serverless inference
• SageMaker transformer for batch inference
• Running Inference with SageMaker Hosting
• Inference with real-time endpoints
• Inference with serverless endpoints
• Inference with Batch Transform
• Adding a SageMaker Elastic Inference (EI) accelerator
• Conclusion
• Points to Remember

Adding Intelligence with Sensory Cognition

• Introduction
• Introducing AWS AI services
• Adding sensory cognition to your applications
• Conclusion
• Points to Remember

AI for Industrial Automation

• Introduction
• Overview of AI for Industrial Automation
• Cost of Poor Quality or COPQ
• Quality Control with Amazon Lookout for Vision
• Predictive Analytics with Amazon Lookout for Equipment
• Conclusion
• Points to Remember

Operationalized Model Assembly (MLOps and Best Practices)

• Introduction
• Lesson Scenario
• MLOps Defined
• Orchestration Options
• Phase Discrimination
• Best Practices using the AWS Well-Architected Lens for Machine Learning
• Conclusion