Reproduce “CARM: Conformal Adaptive Risk Masking for Reliable Scientific Super-Resolution” 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
Guangtao Zhang, Haonan Wu, Junxuan Gao, Zhengyang Liu, Zhaoxi Zeng, Zhihao Li, Jianguo Huang, Zhiqiang Liu
CARM improves reliability in scientific super-resolution by adapting the threshold that turns uncertainty maps into trust masks, while keeping the predictor fixed. It combines threshold proposals with different update speeds and uses delayed feedback to revise them. On a daily South China Sea chlorophyll-a stream, it lowers risk and violations versus static calibration, while retaining less area; individual violations remain possible.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.