Research map / Learning theory and algorithms / Prediction and decision losses
High-dimensional Additive Gaussian Processes under Monotonicity Constraints
Andrés F López-Lopera, Francois Bachoc, Olivier Roustant
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 |
|---|---|---|---|
| additive structure vs non-additive cGP (5D) | additive cGP · 5D monotonic example | Q2 (%): absolute improvement of 1.3% | Improved |
| additive structure vs non-additive cGP (5D) | additive cGP mode · 5D monotonic example | CPU time: 28.9s -> 6s | Improved |
| add monotonicity constraints (cGP mode vs GP mean) | additive GP · d=10 | Q2 (%): 82.3 -> 83.8 | Improved |
| add monotonicity constraints (cGP mean vs GP mean) | additive GP · d=10 | Q2 (%): 82.3 -> 88.1 | Improved |
| add monotonicity constraints (cGP mean vs GP mean) | additive GP · d=1000 | Q2 (%): 92.6 -> 95.1 | Improved |
| use cGP mean instead of cGP mode | additive cGP · d=1000 | CPU time (s): 262.4 -> 10454.3 | Hurt |
| use cGP mean instead of cGP mode | additive cGP · d=1000 | Q2 (%): 94.6 -> 95.1 | Improved |
| increase dimension from d=10 to d=1000 | additive cGP mode · monotonic example | Q2 (%): 83.8 -> 94.6 | Improved |
| additive MaxMod mode vs non-additive cGP with same knots | additive cGP + MaxMod · D=10, d=5 | Q2 (%): 91.4 -> 99.8 | Improved |
| additive MaxMod mode vs non-additive cGP with same knots | additive cGP + MaxMod · D=20, d=5 | Q2 (%): 96.0 -> 99.7 | Improved |
| additive MaxMod mode vs non-additive cGP with same knots | additive cGP + MaxMod · D=10, d=2 | Q2 (%): 99.5 -> 99.8 | Improved |
| MaxMod on flood study (Vienne river) | MaxMod · flood study, n=2d=74 | Q2 | Improved |
Closest papers by meaning
- Robust and Computation-Aware Gaussian Processes (NeurIPS 2025 · similarity 0.85)
- Statistical Guarantees for High-Dimensional Stochastic Gradient Descent (NeurIPS 2025 · similarity 0.85)
- Sparse Gaussian Processes: Structured Approximations and Power-EP Revisited (NeurIPS 2025 · similarity 0.85)
- Vanilla Bayesian Optimization Performs Great in High Dimensions (ICML 2024 · similarity 0.85)
- Relaxing the Additivity Constraints in Decentralized No-Regret High-Dimensional Bayesian Optimization (ICLR 2024 · similarity 0.84)
- High-probability complexity bounds for stochastic non-convex minimax optimization (NeurIPS 2024 · similarity 0.84)
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.
