Reproduce “Learning Placeholders for Open-Set Recognition” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Da-Wei Zhou, Han-Jia Ye, De-Chuan Zhan
PROSER addresses open-set recognition by classifying known classes while rejecting unfamiliar inputs. It adds dummy classifier outputs to learn an adaptive unknown score and mixes hidden representations from known classes, labeling those mixtures unknown during training. On CIFAR-10 with LSUN-crop outliers, it achieves 86.7 macro F1 versus 75.1 for GFROSR. Novel categories remain an aggregated unknown label, not individually identified.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.