Reproduce “A randomized algorithm to solve reduced rank operator regression” 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
Giacomo Turri, Vladimir Kostic, Pietro Novelli, Massimiliano Pontil
The paper introduces Randomized Reduced Rank Regression (R⁴), which fits low-rank operators for vector-valued regression using random sketches, power iterations, and oversampling instead of large generalized eigenvalue problems. Its theory bounds expected regularized empirical-risk gap over sketch randomness, not population risk. On the NSD comparison, R⁴ scores 0.2010 versus R³’s 0.2015, with much shorter plotted fit times.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.