Reproduce “Energy-Based Models for Functional Data using Path Measure Tilting” 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
Jen Ning Lim, Sebastian Vollmer, Lorenz Wolf, Andrew Duncan
F-EBM models functional data as queryable functions, using a kernel-weighted spectral basis and an energy-based prior to represent behavior between observed locations. This lets it evaluate predictions on grids different from the input grid; a FashionMNIST example conditions 100×100 outputs on 14×14 images. On Melbourne, its two-sample-test power is 0.09 versus 0.32 for π-VAE, limited to reported datasets.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.