Reproduce “Random Features for Kernel Approximation: A Survey on Algorithms, Theory, and Beyond” 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
Fanghui Liu, Xiaolin Huang, Yudong Chen, Johan A.K. Suykens
The authors survey random feature methods that approximate kernel comparisons with explicit vectors, letting linear models avoid costly full kernel matrices. For shift-invariant kernels, random Fourier features map inputs using sampled frequencies. Better kernel approximation does not necessarily improve prediction: test error shows double descent in two regression examples. Under stated conditions, leverage-score sampling requires fewer features theoretically.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.