Reproduce “Physically Interpretable Representation Learning with Gaussian Mixture Variational AutoEncoder (GM-VAE)” 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
Tiffany Fan, Murray Cutforth, Marta D'Elia, Alexandre Cortiella, Alireza Doostan, Eric Darve
The GM-VAE organizes 2,950 Schlieren images from 118 laser-ignition experiments into a representation using Gaussian-mixture clusters and alternating updates to learn image reconstructions and cluster memberships. A graph-spectral score tests whether withheld physical measurements vary smoothly across the representation. The embedding shows distinct paths for direct and indirect ignition and dense clusters for non-ignition in this dataset, not proof of mechanisms.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.