Models from experiments referenced in the paper "Training CNNs with Low-Rank Filters for Efficient Image Classification", https://arxiv.org/abs/1511.06744 Model names differ from those in the paper,
Corresponding to the classification accuracy of the sparsest median performing penalized SVM (see Fig 1) for each classifier area under the ROC curve (AUC), sensitivity (SN = TPR, true positive rate),
Values are differences in proportion of correct classification between the models with weight and BMD in each category. Negative values showed that the model with BMD performed better than the model w