Reproduce “TaskNorm: Rethinking Batch Normalization for Meta-Learning” 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
John Bronskill, Jonathan Gordon, James Requeima, Sebastian Nowozin, Richard E. Turner
TaskNorm normalizes few-shot predictions by blending statistics from labeled task examples with statistics from the individual image, using a learned weight based on task size. This lets it classify each query without other target images. In CNAPS on Meta-Dataset, TaskNorm-I ranks first on 11 of 13 datasets, with mean rank 2.38 versus 3.92 for transductive batch normalization; results vary by setting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.