Deep Learning with Keras

Step 5 — testing Theano, TensorFlow, and Keras

Now let's test the environment. First let's look at how to define the sigmoid function in Theano. As you see, it is very simple; we just write the mathematical formula and compute the function element-wise on a matrix. Just run the Python Shell and write the code as shown in the following screenshot to get the result:

So, Theano works. Let's test TensorFlow by simply importing the MNIST dataset as shown in the following screenshot. We have already seen, in Chapter 1, Neural Networks Foundations, a few working examples of the Keras network: