Win $20,000. Help build the future of education. Answer the call. Learn more

Kubernetes with OpenShift 101: Exercises to enhance your apps with machine learning

Machine learning dependencies are a hassle, whether ensuring that the right versions are installed on all systems or that versions of dependencies in your projects are still compatible with the version on your cloud-based system when you deploy. But with containers, you can create a clean, virtual environment to set up and train your neural networks in. Then, you can deploy the neural networks at scale with the exact same environment. To try it yourself, these exercises start with a “Hello World” machine learning app. You build, deploy, and train your neural network, and then deploy it to your local Red Hat OpenShift environment.

In the previous exercises in Kubernetes with OpenShift 101 and Kubernetes with OpenShift 101 Node-RED, you got an introduction to Minishift, a Node.js web server, and running Node-RED on OpenShift.

This tutorial can help you understand how to deploy and manage a machine learning app on Minishift and Red Hat OpenShift on IBM Cloud. When you complete the exercises, you will know how to do the following tasks:

  • Recognize handwritten digits with Keras+, the MNIST dataset
  • Build a neural network
  • Deploy the neural network
  • Deploy the app to Minishift
  • Train the neural network