Current perceptual similarity metrics operate at the level of pixels and patches.

In Figure 1, we compare our LMC framework with the baseline Softmax, and present qualitative results on the TinyImageNet dataset. Below, we discuss them in more detail. AUROC is a widely-used threshold-independent evaluation metric. Both authors contributed equally to the work. Before entering the inference process, similar to our framework, Softmax also pre-stores certain CLIP and DINO features to make the inference process more efficient.

A Rh s s is the attention matrix, where rowirepresents the attention coefficients of each tokenintheinput with respect tothetokeni.

Similarly, self-supervised representation learning (SSL) is rapidly replacing supervised learning as the de-facto pretraining strategy for deep networks, due to improved scalability (unlabeled data is easier to collect) and generality (domain-specific SSL is often preferable to one-fits-all ImageNet pretraining [16,17]).

We partition the domainΩ into a finite set of subregions, called cells or elements.

The detailed spectral clustering9 algorithm is shown in Alg. 1. This spectral clustering algorithm should be inserted into line 7 of10 Alg.1inourmainpaper.11 Interestingly, these two feature clustering strategies lead to similar data16 selection performance on PASCALVOC [7] object detection task. In this part, we pay attention to the effect of pretraining on the final performance of FreeSel. Randaugment: Practical automated data124 augmentation with areduced search space.