Max-Pooling Dropout for Regularization of Convolutional Neural Networks
2015·Arxiv