Reproduce “Sparse Algorithms for Markovian Gaussian Processes” 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
William J. Wilkinson, Arno Solin, Vincent Adam
The paper extends sparse Markovian Gaussian processes to site-based inference for non-Gaussian observations. It approximates each observation with a local Gaussian site and uses inducing-state Markov structure to combine sites efficiently, deriving methods related to power expectation propagation, conjugate-computation variational inference, posterior linearisation, and the extended Kalman smoother. On Audio, its PEP method reports normalized NLPD −1.326 ± 0.01 versus 0.681 ± 0.03 for CVI; performance varies by task.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.