IJCNN 2017 Explainability of Learning Machines