attacking_neural_networks Implementation of l2 attack based on Calini&Wagner's paper: https://arxiv.org/abs/1608.04644