Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
The Multivariate Generalised von Mises Distribution: Inference and Applications
Alexandre Navarro, Jes Frellsen, Richard Turner
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| model circular outputs directly with mGvM regression rather than 2D-GP sine/cosine regression | circular regression · Toy, Uber, Tides, Protein, Yeast validation | log likelihood: Toy: 8.28\E{2} -> 2.02\E{4}; Uber: -2.83\E{2} -> 3.29\E{4}; Tides: -8.41\E{1} -> 1.25\E{4}; Protein: 1.28\E{5} -> 1.42\E | Improved |
| model angles with mGvM regression rather than a non-circular 1D-GP | circular regression · Toy, Uber, Tides, Protein, Yeast validation | log likelihood: Toy: -1.62\E{3} -> 2.02\E{4}; Uber: -1.49\E{3} -> 3.29\E{4}; Tides: -6.46\E{4} -> 1.25\E{4}; Protein: -3.34\E{5} -> 1.42 | Improved |
| use cPCA point estimates instead of cPCA variational Bayes with ARD priors (ablation) | cPCA · Motion Capture denoising | SNR (dB): 2.5 noise: 24.6 -> 29.6; 5 noise: 21.9 -> 23.5; 10 noise: 17.6 -> 17.6 | Improved |
| use cPCA point estimates instead of cPCA variational Bayes with ARD priors (ablation) | cPCA · Robot denoising | SNR (dB): 2.5 noise: 33.2 -> 33.5; 5 noise: 29.8 -> 30.0; 10 noise: 24.8 -> 24.9 | Improved |
| use a variational Bayes cPCA model with ARD priors rather than PPCA | cPCA dimensionality reduction · Motion Capture denoising | SNR (dB): 2.5 noise: 23.6 -> 24.6; 5 noise: 20.9 -> 21.9; 10 noise: 17.2 -> 17.6 | Improved |
| use cPCA point estimates rather than PPCA | cPCA dimensionality reduction · Robot denoising | SNR (dB): 2.5 noise: 22.3 -> 33.5; 5 noise: 21.8 -> 30.0; 10 noise: 20.5 -> 24.9 | Improved |
| use variational inference instead of Gibbs sampling | mGvM inference · Bivariate GvM | runtime: \SI{3.1}{\s} -> \SI{0.02}{\s} | Improved |
Closest papers by meaning
- Domain Invariant Learning for Gaussian Processes and Bayesian Exploration (AAAI 2024 · similarity 0.84)
- Sharp Generalization for Nonparametric Regression by Over-Parameterized Neural Networks: A Distribution-Free Analysis in Spherical Covariate (ICML 2025 · similarity 0.82)
- Learning Latent Variable Models via Jarzynski-adjusted Langevin Algorithm (NeurIPS 2025 · similarity 0.82)
- Maximum Likelihood Estimation is All You Need for Well-Specified Covariate Shift (ICLR 2024 · similarity 0.82)
- Momentum Particle Maximum Likelihood (ICML 2024 · similarity 0.82)
- Preventing Model Collapse in Gaussian Process Latent Variable Models (ICML 2024 · similarity 0.81)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
