A convolutional autoencoder for image data, with options for a configurable number of layers, dropout, and weight regularization.