PMAL: Open Set Recognition via Robust Prototype Mining

---

Abstract

Open Set Recognition (OSR) has been an emerging topic. Besides recognizing predefined classes, the system needs to reject the unknowns. Prototype learning is a potential manner to handle the problem, as its ability to improve intra-class compactness of representations is much needed in discrimination between the known and the unknowns. In this work, we propose a novel Prototype Mining And Learning (PMAL) framework. It has a prototype mining mechanism before the phase of optimizing embedding space, explicitly considering two crucial properties, namely high-quality and diversity of the prototype set. Concretely, a set of high-quality candidates are firstly extracted from training samples based on data uncertainty learning, avoiding the interference from unexpected noise. Considering the multifarious appearance of objects even in a single category, a diversity-based strategy for prototype set filtering is proposed. Accordingly, the embedding space can be better optimized to discriminate therein the predefined classes and between known and unknowns. Extensive experiments verify the two good characteristics (i.e., high-quality and diversity) embraced in prototype mining, and show the remarkable performance of the proposed framework compared to state-of-the-arts. [Paper]

---

Highlights Contributions

❃ Different from the common usage of implicitly learnable prototypes, we pay more attention on choosing prototypes with explicit criteria for OSR tasks. We point out the two important attributes of prototypes, namely the high-quality and diversity.

❃ We design a OSR framework by prototype mining and learning. In the prototype mining phase, the above two key attributes are taken into consideration. In the embedding learning phase, with the chosen prototypes as fixed anchors for each class, a better embedding space is learned, without any sophisticated skills for convergence.

❃ Extensive experiments on multiple OSR benchmarks show that our method is powerful to discriminate the known and unknowns, surpassing the state-of-the-art performance by a large margin, especially in complicated large-scale tasks.

---

Recommended Citations

If you find our work is helpful to your research, please feel free to cite us:

@inproceedings{jing@aaai2022,
  author    = {Jing Lu and
               Yunlu Xu and
			   Hao Li and
               Zhanzhan Cheng and
               Yi Niu and
			   },
  title     = {{PMAL}: Open Set Recognition via Robust Prototype Mining},
  booktitle = {AAAI},
  year      = {2022},
}