Dropout with Tabu Strategy for Regularizing Deep Neural Networks