In the problems of image retrieval and few-shot classification, the mainstream approaches focus on learning a better feature representation.

Such assumptions are necessary for exact recovery to be information-theoretically possible.

We improve the theoretical and empirical performance of neural-network(NN)- based active learning algorithms for the non-parametric streaming setting.

As training data become larger, distributed training has been broadly adopted in machine learning tasks.

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