Back to Glossary
Autoencoder
Models & Architectures
Unsupervised model for compression and reconstruction.
Autoencoders learn a latent representation by reconstructing their input data.
- Applications: Dimensionality reduction, denoising, pretraining.
- Variants: Convolutional, denoising, sparse, variational (VAE).
- Considerations: Latent space quality, reconstruction loss, overfitting.