Auto-Encoder

An Auto-Encoder is a neural network trained to reconstruct its own input. It has two parts: an encoder, which compresses the input into a lower-dimensional latent representation (the bottleneck), and a decoder, which reconstructs the input from that representation. See Encoder-Decoder.

It's trained with a reconstruction loss, such as mean squared error between the input and output. Because the bottleneck is smaller than the input, the network is forced to learn a compressed representation that keeps the most important information.

Auto-encoders are used for Dimensionality Reduction, denoising and representation learning. The Variational Auto-Encoder is a probabilistic version that can also be used to generate new samples.