Introduction

As a data scientist, you need to train numerous models to identify the right combination of data in conjunction with hyperparameters to optimize the performance of your neural networks. You want to perform more experiments faster. You want to train deeper networks and explore broader hyperparameters spaces. IBM Watson Machine Learning accelerates this iterative cycle by simplifying the process to train models in parallel with an on-demand GPU compute cluster.

Here's how to get started:

1. Set up your environment for IBM Watson Machine Learning.

Note: When provisioning an instance of IBM Watson Machine Learning to use in training Deep Learning models please be sure to select the Region as US South. Deep learning functionality is not yet available in other regions. You can still use the London region for deployment and scoring of models that are originally created in that region.

2. There are several ways to train models

Use one of the following methods to train your model:

For details, follow the CLI tutorial using Tensorflow.

3. Configure each training run

IBM Watson Machine Learning allows you to rapidly conduct deep learning iterations by submitting multiple training runs that can be queued for training. A training run consists of the following parts:

4. Upload training data to the cloud

Before you can start training your neural networks, you first need to move your data into IBM Cloud. To do this, upload your training data to an IBM Cloud Object Storage service instance. When you're done training, the output from your training runs is written to your IBM Cloud Object Storage so that you can drag files to your desktop.

5. Start training

After setting up your training runs, use the Python client or the CLI to submit your training runs to IBM Watson Machine Learning. IBM Watson Machine Learning packages each of your training runs and allocates them to a Kubernetes container with the requested resources and deep learning framework. Training runs are executed in parallel depending on the GPU resources available to your account level.

6. Deploy to a REST endpoint and obtain predictions

Once you're trained and selected an optimal model, you're ready to deploy your model as a REST endpoint for access by your application.

Next Steps

The fastest way to learn is by example:

Learn more about LSTM networks from this Developer Works article: Introducing deep learning and long-short term memory networks.