Autoencoder based image compression: can the learning be quantization independent?

Open in new window