Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian Data Sketching for Varying Coefficient Regression Models
Rajarshi Guhaniyogi, Laura Baracaldo, Sudipto Banerjee
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 |
|---|---|---|---|
| Compress data with Gaussian sketch (geoS) instead of fitting the uncompressed model | Spatially varying coefficient model · Simulation N=100000 | Computation efficiency: 0.981 -> 5.9081 | Improved |
| Compress data with Gaussian sketch (geoS) instead of fitting the uncompressed model | Spatially varying coefficient model · Simulation N=5000 | Computation efficiency: 0.6298 -> 2.1165 | Improved |
| Compress data with Gaussian sketch (geoS) instead of fitting the uncompressed model | Spatially varying coefficient model · Simulation N=10000 | Computation efficiency: 0.4981 -> 2.2368 | Improved |
| Compress data with Gaussian sketch (geoS) instead of fitting the uncompressed model | Spatially varying coefficient model · Simulation N=5000 | MSE (SVC): 0.0109 -> 0.0335 | Hurt |
| Compress data with Gaussian sketch (geoS) instead of fitting the uncompressed model | Spatially varying coefficient model · Simulation N=100000 | MSPE: 0.112 -> 0.1242 | Hurt |
| Compress data with Gaussian sketch (geoS) instead of fitting the uncompressed model | Spatially varying coefficient model · Simulation N=10000 | 95% CI coverage: 0.9500 -> 0.9253 | No clear effect |
| Use geoS instead of sparse GP predictive process | Varying intercept model, N=5000 · Simulation N=5000 vs sparse GP (M=710) | Time (secs): 5190.77 -> 17.47 | Improved |
| Use geoS instead of sparse GP predictive process | Varying intercept model, N=5000 · Simulation N=5000 vs sparse GP (M=142) | MSPE: 0.128 -> 0.146 | Hurt |
| Use geoS on real MODIS NDVI data instead of uncompressed model | Spatially varying coefficient model · MODIS NDVI vegetation (N=113000, M=2300) | Computation efficiency: 0.46901 -> 3.5424 | Improved |
| Use geoS on real MODIS NDVI data instead of uncompressed model | Spatially varying coefficient model · MODIS NDVI vegetation (N=113000, M=2300) | MSPE: 0.00276 -> 0.00327 | No clear effect |
| Increase compression dimension M = k*sqrt(N) up to k about 10 (ablation) | geoS · Simulation N=100000 | MSPE: k=1 to 20; MSPE decline slows near k~10 | Improved |
| Increase compression dimension M = k*sqrt(N) up to k about 10 (ablation) | geoS · Simulation N=100000 | 95% predictive interval length: stabilizes around k~10 | Improved |
Closest papers by meaning
- Distributed Bayesian Varying Coefficient Modeling Using a Gaussian Process Prior (JMLR 2022 · similarity 0.91)
- Sketching in High-Dimensional Regression With Big Data Using Gaussian Scale Mixture Priors (JMLR 2025 · similarity 0.87)
- Bayesian Additive Distribution Regression (NeurIPS 2026 · similarity 0.85)
- Inflationary Flows: Calibrated Bayesian Inference with Diffusion-Based Models (NeurIPS 2024 · similarity 0.84)
- Quasi-Bayes meets Vines (NeurIPS 2024 · similarity 0.84)
- Asymptotically Log-Optimal Bayes-Assisted Confidence Sequences for Bounded Means (NeurIPS 2026 · similarity 0.83)
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.
