Autoencoder
The nervous network that learns efficiently by compressing the input and learning to recreate again.
The autocoder (autoencoder) is an unsupervised nervous network architecture that learns to recreate the original input from this compressed representation as faithful as possible. The structure consists of two main parts: a encoder (encoder), compresses the input data into a low-dimensional cover space (latent space) called "darboğaz"; a decoder (decoder) tries to rebuild the original data from this compressed representation. The model is trained to minimize the difference between the inlet and the outlet, because it does not require labeled data in this process, because the model entered "right answer".
The importance of autocoders comes from pushing the model of the bottleneck to learn the most important features of the data; it can be used in many areas such as compressed representation, size reduction, noise removal (denoising autoencoder), anomaly detection and recommendation systems. Varel autocoders (VAE) with a potential version of autocoders can produce new, realistic data by sampling from the covering space, which has made them one of the leading architecture of the manufacturer artificial intelligence.
The importance of autocoders comes from pushing the model of the bottleneck to learn the most important features of the data; it can be used in many areas such as compressed representation, size reduction, noise removal (denoising autoencoder), anomaly detection and recommendation systems. Varel autocoders (VAE) with a potential version of autocoders can produce new, realistic data by sampling from the covering space, which has made them one of the leading architecture of the manufacturer artificial intelligence.