Reproduce “An Instance-Dependent Simulation Framework for Learning with Label Noise” 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
Keren Gu, Xander Masotto, Vandana Bachani, Balaji Lakshminarayanan, Jack Nikodem, Dong Yin
The authors simulate instance-dependent label noise by training diverse rater models on clean data and using their predictions to label other examples. Rater features also support LQM, which estimates corrected labels using a paired subset with cleaner labels. On CIFAR10-H, synthetic labels more closely matched human label distributions than random flips under this comparison.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.