Tutorials demonstrating Watson Machine Learning using the MNIST database

This topic lists tutorials that demonstrate IBM Watson Machine Learning interfaces and deep learning features, as well as IBM Watson Studio tools. All of these tutorials tackle the same challenge: to build a machine learning model or simple neural network that recognizes handwritten digits, using the MNIST data set as training data.

Running tutorials for the legacy v1 Machine Learning service

During the migration period, you can still run tutorials and examples associated with a legacy v1 Watson Machine Learning service instance. However, note the following requirements and restrictions:

  • You must use a v1 service instance and associated credentials to run a v1 sample or example. Follow the authentication steps to authenticate with deprecated samples.
  • Lite users can use existing v1 service credentials, but cannot create new credentials.
  • Standard and Professional users can use existing v1 service credentials and can also create new v1 service credentials. The Credentials page was removed from the IBM Cloud Services catalog, so follow the steps in Generating legacy Watson Machine Learning credentials to create new credentials using the IBM Cloud CLI.

About the sample data

The Modified National Institute of Standards and Technology (MNIST) database contains images of handwritten digits. It is popular for machine learning and deep learning exploration and study.

See also:

 

About the sample code

You don’t have to build a model from scratch for any of these tutorials. The model builder tutorial and the flow editor tutorial use no code at all. The rest of the tutorials use sample model-building code:

tf-model.zip and tf-model-hpo.zip both contain two files:

  • convolutional_network.py - Model-building Python code
  • input_data.py - A "helper" file for reading the MNIST data files

 

About the tutorials

All of these tutorials demonstrate basic Watson Machine Learning functionality:

  • Train a model using Watson Machine Learning specialized, high-performance infrastructure
  • Deploy a model to IBM Cloud
  • Serve API calls (to use the deployed model to classify new images)

Table 1 lists the tutorials:

  • For each tutorial, Table 1 shows which tools and interfaces are demonstrated, which deep learning features are demonstrated, and what you need to download or install on your local computer to complete the tutorial.

  • The tutorials are presented in Table 1 in ascending order of complexity, where complexity is based on how many tools (if any) need to be installed, how many interfaces are used, how many files need to be downloaded, and how long it takes to complete the tutorial. Complexity is given on a scale from "1" (least complex) to "4" (most complex).

Table 1. Watson Machine Learning tutorials using MNIST
Tutorial Interfaces used Features demonstrated Local download or install
MNIST experiment builder tutorial

 

Complexity: 2

  • Cloud Object Storage GUI
  • Watson Studio project
  • Experiment builder in Watson Studio
  • Using GPUs
  • Framework: TensorFlow
MNIST experiment builder HPO tutorial

 

Complexity: 2

  • Cloud Object Storage GUI
  • Watson Studio project
  • Experiment builder in Watson Studio
  • Using GPUs
  • Hyperparameter optimization
  • Framework: TensorFlow
MNIST Python client tutorial

 

Complexity: 2

  • Cloud Object Storage GUI
  • Watson Studio project
  • Watson Machine Learning Python client
  • Notebook in Watson Studio
  • Using GPUs
  • Framework: TensorFlow
MNIST CLI tutorial

 

Complexity: 3

  • Cloud Object Storage GUI
  • Watson Machine Learning command line interface
  • Using GPUs
  • Framework: TensorFlow
MNIST CLI HPO tutorial

 

Complexity: 3

  • Cloud Object Storage GUI
  • AWS command line interface (for Cloud Object Storage)
  • Watson Machine Learning command line interface
  • Using GPUs
  • Hyperparameter optimization
  • Framework: TensorFlow