Reproduce “Exponential Convergence of CAVI for Bayesian PCA” 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
Arghya Datta, Philippe Gagnon, Florian Maire
The authors analyze convergence of coordinate ascent variational inference for Bayesian probabilistic PCA. They show that with one component, normalized updates follow power iteration and converge exponentially toward PCA directions under conditions. For any number of components, they prove exponential convergence only locally, under assumptions and near a stationary point. This guarantee concerns fitting, not runtime or a global optimum.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.