Reproduce “Learning Population Dynamics with Variational Doob-Wasserstein Flow” 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
Weizhu Qian
VDWF addresses single-cell trajectory inference from sparse snapshots and changing populations. It constructs data-shaped paths guided by observed endpoints, and uses WFR interpolation to model mass changes without trajectory simulation during training. On normalized PCA-5 CITE-seq, it reports average hold-out Wasserstein-1 distance 0.735 ± 0.035 versus 0.750 ± 0.033 for WFR-FM, a modest advantage in this setting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.