Autoencoders
What is autoencoder
- An autoencoder is a neural network architecture designed to learn compressed (lower-dimensional) representations of data in an unsupervised manner
- Unlike traditional supervised neural networks (like CNNs or RNNs), autoencoders are trained to reproduce their own input.
- It is not a specific type like CNN or RNN, but rather a design architecture built using standard layers
Key Components
- Encoder: Maps input data to a lower-dimensional latent space
- Latent Representation: The compressed code containing key features
- Decoder: Reconstructs input from the latent code