What is the difference between this work and CVPR21 paper?

[7Anonym]

I came across an article titled Learning placeholders for open-set recognition, which was published in CVPR 2021. I find these works similar in the title and the techniques used for building placeholders.

Could the authors please provide clarification on the differences between these works?

[zaiquanyang]

I came across an article titled Learning placeholders for open-set recognition, which was published in CVPR 2021. I find these works similar in the title and the techniques used for building placeholders.

Could the authors please provide clarification on the differences between these works?

Thanks for your attention for the work. LPL utilizes 'placeholders' to illustrate the motivation, which looks similar but there are some differences in my opinion: PROSER leverages the 'placeholders' to learn the extra dummy classifiers for rejecting the novel classes but the LPL does not explicitly introduced more classifiers and focuses on the diversified distribution of 'placeholders' and cross-modal alignment. Besides, the another key of LPL is to divide the ZSL into representation learning and discriminant learning two stages. Hope the above can answer your question. If any, please contact my email for timely reply.