In the sections to follow, we will show how to install Keras on multiple platforms.

First, we install the numpy package, which provides support for large, multidimensional arrays and matrices as well as high-level mathematical functions. Then we install scipy, a library used for scientific computation. After that, it might be appropriate to install scikit-learn, a package considered the Python Swiss army knife for machine learning. In this case, we will use it for data exploration. Optionally, it could be useful to install pillow, a library useful for image processing, and h5py, a library useful for data serialization used by Keras for model saving. A single command line is enough for installing...

Keras has a very minimalist configuration file. Let's load it with a vi session. The parameters are very simple:

The image_dim_ordering of th value gives you a somewhat non-intuitive dimension ordering for images (depth, width, and height), instead of (width, height, and depth), for tf. The following are the default parameters in my machine:

One of the easiest ways to get started with TensorFlow and Keras is running in a Docker container. A convenient solution is to use a predefined Docker image for deep learning created by the community that contains all the popular DL frameworks (TensorFlow, Theano, Torch, Caffe, and so on). Refer to the GitHub repository at https://github.com/saiprashanths/dl-docker for the code files. Assuming that you already have Docker up and running (for more information, refer to https://www.docker.com/products/overview), installing it is pretty simple and is shown as follows:

The following screenshot, says something like, after getting the image from Git, we build the Docker image:

In this screenshot, we see how to run it:

From within the container, it is possible to activate support for Jupyter Notebooks (for more information, refer to http://jupyter.org/):

Access it directly...

Installing Keras on Google Cloud is very simple. First, we can install Google Cloud (for the downloadable file, refer to https://cloud.google.com/sdk/), a command-line interface for Google Cloud Platform; then we can use CloudML, a managed service that enables us to easily build machine, learning models with TensorFlow. Before using Keras, let's use Google Cloud with TensorFlow to train an MNIST example available on GitHub. The code is local and training happens in the cloud:

In the following screenshot, you can see how to run a training session:

We can use TensorBoard to show how cross-entropy decreases across iterations:

In the next screenshot, we see the graph of cross-entropy:

Now, if we want to use Keras on the top of TensorFlow, we simply download the Keras source from PyPI (for the downloadable file, refer to https://pypi.Python.org/pypi...

Installing TensorFlow and Keras on Amazon is very simple. Indeed, it is possible to use a prebuilt AMI named TFAMI.v3 that is open and free (for more information, refer to https://github.com/ritchieng/tensorflow-aws-ami), shown as follows:

This AMI runs TensorFlow in less than five minutes and supports TensorFlow, Keras, OpenAI Gym, and all dependencies. As of January 2017, it supports the following:

• TensorFlow 0.12
• Keras 1.1.0
• TensorLayer 1.2.7
• CUDA 8.0
• CuDNN 5.1
• Python 2.7
• Ubuntu 16.04

In addition, TFAMI.v3 works on P2 computing instances (for more information, refer to https://aws.amazon.com/ec2/instance-types/#p2), as shown in the following screenshot:

Some features of P2 instances are as follows:

• Intel Xeon E5-2686v4 (Broadwell) processors
• NVIDIA K80 GPUs, each with 2,496 parallel cores and 12 GB of GPU memory
• Supports peer-to-peer GPU communication
• Provides...

One way to install Keras on Azure is to install the support for Docker and then get a containerized version of TensorFlow plus Keras. Online, it is also possible to find a detailed set of instructions on how to install Keras and TensorFlow with Docker, but this is essentially what we have seen already in a previous section (for more information, refer to https://blogs.msdn.microsoft.com/uk_faculty_connection/2016/09/26/tensorflow-on-docker-with-microsoft-azure/).

If you use Theano as the only backend, then Keras can run with just a click by loading a pre-built package available on Cortana Intelligence Gallery (for more information, refer to https://gallery.cortanaintelligence.com/Experiment/Theano-Keras-1).
The following sample shows how to import Theano and Keras into Azure ML directly as a ZIP file and use them in the Execute Python Script module. This example is due to...