Reproduce “Learning a Neural-network-based Representation 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

Mehadi Hassen, Philip K. Chan

The authors propose an open-set malware classifier that trains representations to compact known classes and separate their means, then uses distance to a class mean and a threshold to flag unknowns. On simulated class-withholding tests, II+CE achieves a macro-averaged F-score of 0.889 on Microsoft Malware Challenge versus 0.868 for OpenMax. These results do not establish real-world deployment performance.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.