Machine Learning Design Patterns

Index

A

• ablation, Data augmentation
• AdaBoost, Boosting, Increased training and design time
• AdaNet, Other ensemble methods
• AI Platform Notebooks, Explanations from deployed models
• AI Platform Pipelines, Scheduled retraining, Running the pipeline on Cloud AI Platform
• AI Platform Prediction, Data and Model Tooling, Running the pipeline on Cloud AI Platform, Other serverless versioning tools, After training
• AI Platform pusher component, Running the pipeline on Cloud AI Platform
• AI Platform Training, Data and Model Tooling, Fully managed hyperparameter tuning
• AI readiness, AI Readiness-Transformational phase: Fully automated processes
• Alexander, Christopher, What Are Design Patterns?-What Are Design Patterns?
• all-reduce algorithm, Synchronous training
• anomaly detection, Anomaly detection-Choosing a model architecture, Handling many predictions in near real time, Fraud and Anomaly Detection
• Apache Airflow, Solution, Apache Airflow and Kubeflow Pipelines
• Apache Beam, Batch and stream pipelines, Efficient transformations with tf.transform, Solution-Reduce computational overhead, Solution, Alternative implementations
• Apache Flink, Solution, Alternative implementations
• Apache Spark, Batch and stream pipelines, Solution, Alternative implementations
• Apigee, Model versioning with a managed service
• application-specific integrated circuit (see ASIC)
• ARIMA, Problem Representation Design Patterns, Solution
• arrays, Array of numbers-Array of numbers
• ASIC, ASICs for better performance at lower cost, Phase 1: Building the ...