Reproduce “Multi-Label Prediction via Compressed Sensing” 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
Daniel Hsu, Sham M. Kakade, John Langford, Tong Zhang
The authors introduce a method for multi-label prediction that reduces the number of predictors by compressing labels into measurements, predicting those measurements, and reconstructing a sparse label vector. On image and text tagging tasks, 300 or 400 measurements gave nearly the same precision-at-k as one-against-all with 1,024. The guarantee depends on sparsity of the conditional mean labels.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.