Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Large-Scale Cox Process Inference using Variational Fourier Features
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use Fourier features instead of RBF inducing points | Cox process inference · Synthetic 1D examples | test set likelihood: less than 1% | No clear effect |
| Use Fourier features instead of RBF inducing points | Cox process inference · Porto taxi pickups, 2D | optimization speed: VFF took about one second per gradient step; RBF+IP took 50 to 500 second per gradient step | Helped |
| Use a constant offset β in the square-root link | VFF Cox process · Porto taxi pickups, 2D | nodal-line artifacts | Helped |
| Remove the constant offset by fixing β=0 (ablation) | VFF Cox process · Porto taxi pickups, 2D | nodal-line artifacts | Hurt |
| Increase the number of Fourier frequencies (ablation) | VFF Cox process · Porto taxi pickups, 2D | rate-function resolution | Helped |
| Use a periodic kernel on the time dimension | VFF spatiotemporal Cox process · Porto taxi pickups, 3D | rate continuity at midnight | Helped |
| Use a periodic time kernel rather than a non-periodic time kernel (ablation) | VFF spatiotemporal Cox process · Porto taxi pickups, 3D | rate continuity at midnight | Hurt |
| Use the mean rate for test-set scoring instead of posterior samples | Variational Cox process inference · Synthetic 1D examples | test set likelihood agreement | No clear effect |
| Use an insufficient number of inducing features (ablation) | VFF Cox process · Synthetic 1D examples | lengthscale estimation | Hurt |
| Use inducing points rather than Fourier features | Cox process inference · Porto taxi pickups, 2D | optimization robustness | Hurt |
Closest papers by meaning
- Spatio-Temporal Variational Gaussian Processes (NeurIPS 2021 · similarity 0.91)
- Bayesian Optimization through Gaussian Cox Process Models for Spatio-temporal Data (ICLR 2024 · similarity 0.90)
- Fast Bayesian Inference for Gaussian Cox Processes via Path Integral Formulation (NeurIPS 2021 · similarity 0.90)
- Efficient Variational Inference for Log-Gaussian Cox Processes via Voronoi Tessellation (NeurIPS 2026 · similarity 0.90)
- Sparse Gaussian Processes: Structured Approximations and Power-EP Revisited (NeurIPS 2025 · similarity 0.88)
- Turbocharging Gaussian Process Inference with Approximate Sketch-and-Project (NeurIPS 2025 · similarity 0.87)
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.