Accelerating TensorFlow with the Google Machine Learning Engine

Learn how to use TensorFlow to build high-performing machine learning apps. Discover how to develop and run applications on the Google Cloud Machine Learning Engine.

Discover how to leverage TensorFlow—an open-source software library for numerical computation—to build high-performing machine learning applications. In this course, instructor Matt Scarpino helps to acquaint you with this exciting tool. Here, he explores the process of developing TensorFlow applications and running them on the Google Cloud Machine Learning (ML) Engine.

Matt kicks off the course by discussing TensorFlow development in detail, starting with basic tensor operations and proceeding to graphs, sessions, variables, and training. He also goes over high-level features like datasets, iterators, and estimators. Next, Matt introduces the Google Cloud Platform (GCP) and its capabilities. He shows how to create a GCP project and access it through the Cloud SDK utility. In addition, he covers Google Cloud Storage, which enables developers to upload data that can be accessed in GCP applications. To wrap up, he steps through how to deploy your TensorFlow applications to the ML Engine.

Introduction

• Welcome
• What you should know
• Using the exercise files

1. Introducing TensorFlow

• Overview and installation
• Getting started
• Running a simple application

2. Fundamentals of TensorFlow Development

• Creating tensors
• Basic tensor operations
• Advanced tensor operations
• Understanding graphs and sessions
• Accessing graphs and sessions in code

3. Training TensorFlow Applications

• Variables and logging
• Using variables in code
• Using optimizers
• Simple optimizer example
• Batches and placeholders
• Linear regression in code: Part 1
• Linear regression in code: Part 2
• TensorBoard
• Using TensorBoard in practice

4. Accessing Data with Datasets

• Datasets and iterators
• Coding with datasets and iterators
• Dataset operations
• Creating datasets from files
• Introducing MNIST images
• Reading MNIST data in code

5. Machine Learning with Estimators

• Understanding estimators
• Describing data with feature columns
• Coding a simple estimator: Part 1
• Coding a simple estimator: Part 2
• Estimators and neural networks
• Coding a DNN estimator: Part 1
• Coding a DNN estimator: Part 2
• Automating estimator operation
• Estimator automation in practice

6. Deploying Estimators to the Machine Learning Engine

• Creating a GCP project
• Installing the Cloud SDK
• Introduction to Google Cloud Storage
• Accessing Cloud Storage in practice
• Machine Learning Engine
• Deploying jobs to ML Engine

Conclusion

• Next steps