An autoencoder is a special kind of feedforward neural network capable of learning efficient encoding of input data. These encodings can be of a lower or higher dimension than the input. Autoencoder is an unsupervised deep learning technique that learns to represent input data into a latent feature space. Autoencoders can be leveraged for multiple applications such as dimensionality reduction, image compression, image denoising, image generation, and feature extraction.

In this chapter, we will cover the following recipes:

• Implementing vanilla autoencoders
• Dimensionality reduction using autoencoders
• Denoising autoencoders
• Changing black and white to color