Using a TensorFlow Neural Network to Style Images
Description
Lab Overview
Neural networks have been used for many applications throughout the deep learning revolution. In this Lab, you will use the AWS Deep Learning AMI using a GPU instance (g4dn.xlarge). You will perform neural style transfers - an algorithm for combining the content of one image with the style of another image. This process involves using convolutional neural networks (CNN). You will then be able to examine the GPU performance and cost associated with the CNN as it runs.
Lab Objectives
Upon completion of this Lab you will be able to:
- Perform neural style transfers using the AWS Deep Learning AMI
- Publish GPU metrics to Amazon CloudWatch
- Examine GPU performance in Amazon CloudWatch
Lab Prerequisites
You should be familiar with:
- Working with Linux on the command-line
- Graphics processing unit (GPU) concepts
- Knowledge of the Python programming language is beneficial, but not required
Lab Environment
Before completing the Lab instructions, the environment will look as follows:
After completing the Lab instructions, the environment should look similar to:
Updates
May 13, 2025 - Updates to use TensorFlow as MXNet has been deprecated
July 17th, 2024 - Updated the instructions to reflect the latest UI
October 30th, 2023 - Resolved an issue that caused the lab to fail to provision on rare occasions
December 5th, 2022 - Added checks to track lab progress
September 8th, 2022 - Recovered the GPU monitoring script functionality
August 31st, 2020 - Updated screenshots for the new EC2 user interface
January 10th, 2019 - Added a validation Lab Step to check the work you perform in the Lab