Reproduce “FiGuRO: Intrinsic Dimension Estimation for Multi-Modal Data” 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
Viktoria Schuster, Sana Tonekaboni, Caroline Uhler
FiGuRO estimates shared and modality-specific intrinsic dimensions during autoencoder training by adjusting low-rank projections according to reconstruction fidelity. In four simulated two-modality datasets at distortion budget lambda equals 0.05, its average absolute rank deviation was 2.9 ± 0.7, compared with 9.5 ± 2.1 for ShIndICA, the best tested baseline. The results do not establish exact recovery for arbitrary data.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.