Two-Level Attention Model of Representation Learning for Fraud Detection | IEEE Journals & Magazine | IEEE Xplore