CPG enables the model to fully capture the intra-category diversity by representing each category with multiple prototypes.

ImageNet-100 is a subset of 100 categories from the coarse-grained ImageNet [13] dataset.

CIFAR-10-L T, CIFAR-100-L T, ImageNet-100-L T, and Places-L T are 5, 80, 50, and 182 respectively. Our default training set of each dataset is summarized in Table 8.

While several previous works have focused on classifying close-set samples and detecting open-set samples during testing, it's still essential to be able to classify unknown subjects as human beings.

Recent advancements have shown promise in applying traditional Semi-Supervised Learning strategies to the task of Generalized Category Discovery (GCD).

We refer to it as theContinuous Category Discovery(CCD) problem, which is significantly more challenging than the static setting.

In this paper we tackle the problem of Generalized Category Discovery (GCD).

TIDA intrinsically: (a) builds hierarchical vMF distributions to cluster samples and discovery taxonomic context by optimizing Eq.