Using Google TensorFlow with Amazon SageMaker
Abhishek Mishra · 2019
This chapter explains how to create models using Google TensorFlow and use Amazon SageMaker to train and deploy these models. TensorFlow is typically used to create different types of artificial neural network models for classification and regression tasks. The chapter provides a high-level summary of artificial neural networks to help one get the general idea of how they work. It then explains how use an Amazon SageMaker notebook instance to train a linear regression model on the notebook instance with the low-level Python API provided by TensorFlow. To better illustrate the process of using Google TensorFlow, the linear regression model is first trained on a single feature of the dataset, and then modified to work with the remaining features. The chapter also explains how use the high-level TensorFlow Estimators API to train a Dense Neural Network-based classifier on the popular Iris flowers dataset.