Reproduce “Scalable and Adaptive Distribution Regression via Random Features” 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
Yinsong Wang, Xiao Liu, Yu Ding
RaDR approximates distribution regression by averaging randomized features within each sample bag, replacing nested instance-level kernel comparisons with compact bag vectors. A-RaDR extends the method by scoring a candidate pool and retaining useful features. On a 41-dataset classification benchmark, its Gaussian-kernel variant achieved average accuracy close to exact KME Gaussian with much lower reported training time. The risk guarantee covers RaDR, not A-RaDR, under stated conditions for kernel ridge regression with a linear distribution kernel.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.