Fig. 1From: Iterative cleaning and learning of big highly-imbalanced fraud data using unsupervised learningAutoencoder visualization of the encoder and decoder componentsBack to article page