A Implementation details A.1 Datasets

For datasets with low/medium number of categories we used CIFAR-10 and CIFAR-100 (Krizhevsky et al., In the finetuning experiments we used the STL-10 dataset (Coates et al., 2011) For datasets with an high number of categories we used the tiny-ImageNet and SlimageNet (Antoniou et al., We use off-the-shelf Pytorch implementations of ResNets as described in the original paper (He et al., 2016). All the methods could fit on a single one of those GPUs. This baseline consists of standard supervised training. It represents an upper bound. When evaluated for the number of augmentations (Appendix B.6) the same strategy adopted in our method (Appendix A.3) has been used to Clustering has been performed at the beginning of each epoch by using the k-means algorithm available in Scikit-learn.