Reproduce “COSMIO: A Benchmark for Cross-Survey Modality Imputation” 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
Dichang Zhang, Yixuan Shao, Jiaqi Deng, Jiali Cui, Haotian Yin, Yuanpeng Liu, Xiang Gao, Zhiqiang Lao and 3 more
COSMIO benchmarks cross-survey image imputation using simulated observations of 115,000 common-galaxy scenes and 14,825 strong-lensing scenes across LSST, Euclid, and Roman. Asyn-Diff, its strongest reported baseline, has average CRPS 0.069 for common-galaxy Roman-target tasks and 0.090 for lens tasks. Imputed-image lens detection also trails ground-truth training; results remain limited to simulated observations.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.