Use public assets on AI Hub to build an ML system

This guide introduces you to the public assets on AI Hub that you can use to build your artificial intelligence (AI) systems. As an example, this guide shows you how to find a Jupyter notebook and open it on Colaboratory (Colab). The example notebook guides you through the process of training two TensorFlow image classification models using the MNIST dataset.

Find and use a Jupyter notebook

Follow these steps to find a notebook that uses the MNIST dataset and then open the notebook in Colab:

  1. Go to AI Hub.
  2. In the Menu > Category pane, click Notebooks. A list of notebooks opens in a new window.
  3. Enter mnist into the search box at the top of the page, to filter the list of notebooks.
  4. Scan the page for a notebook named Classifying Handwritten Digits, and click the name of the notebook. AI Hub displays a description of the notebook and information about how to use it.
  5. In the Use this asset pane, click Open in Colab. The Colab notebook opens in a new browser tab.
  6. Follow the instructions in the Colab notebook to train two TensorFlow image classification models and visualize the results. If you need help with using the notebook interface, click Help in the Colab menu.
  7. Return to AI Hub when you're ready to discover more AI frameworks and tools.

What's next

  • Understand important concepts and terms by reading the introduction to AI Hub.
  • Explore other asset categories on AI Hub, by clicking a category in the left-hand panel and following the prompts to open and use the asset.
  • Get started uploading assets on AI Hub.