Reproduce “Learning to Solve Multiresolution Matrix Factorization by Manifold Optimization and Evolutionary Metaheuristics” 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
Truong Son Hy, Thieu Khang, Risi Kondor
The authors develop a learnable multiresolution matrix factorization method that searches wavelet indices and jointly optimizes orthogonal rotations. It replaces sequential greedy choices with evolutionary index search followed by gradient-based optimization on the Stiefel manifold. Across three tested matrices, it reports lower Frobenius error than original MMF and Nyström at matched active-column counts, though learnable MMF uses larger rotations.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.