In this chapter, we're going to briefly discuss the most common deep learning layers, giving two examples based on Keras. The first one is a deep convolutional network employed to classify the MNIST dataset. The other one is an example of time-series processing using a recurrent network based on Long Short-Term Memory (LSTM) cells. We're also introducing the basic concepts of TensorFlow, giving some concrete examples based on algorithms already discussed in previous chapters.
In particular, we're going to discuss the following:
- Deep learning layers (convolutions, dropout, batch normalization, recurrent)
- An example of a deep convolutional network
- An example of a recurrent (LSTM-based) network
- A brief introduction to TensorFlow with examples of gradient computation, logistic regression, and convolution
