Neural network's architectures can be very different; these configurations are often organized on different layers, the first of which receives the input signals and the last returns the output signals. Usually these networks are identified as feed-forward neural networks.
Feed-forward neural networks, which we intend to illustrate briefly, are well suited to be used for the approximation of functions and for the interpolation.
The following topics are covered in this chapter:
- Introducing feed-forward neural network
- Classification of handwritten digits
- Exploring the MNIST dataset
- Softmax classifier
- How to save and restore a TensorFlow model
- Implementing a five-layer neural network
- ReLU classifier
- Dropout optimization