Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
On the Identifiability and Interpretability of Gaussian Process Models
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 |
|---|---|---|---|
| Add smoother Matérn components to an additive mixture | Single-output GP with Matérn kernels of varying smoothness · Asymptotic fixed-domain setting | smoothness | No clear effect |
| Use an additive mixture instead of its least smooth Matérn component alone | Single-output GP · Asymptotic prediction | MSE | No clear effect |
| Add Matérn 3/2 and 5/2 components to Matérn 1/2 (ablation) | Single-output GP · MNIST image prediction | prediction performance | No clear effect |
| Add Matérn 3/2 and 5/2 components to Matérn 1/2 (ablation) | Single-output GP · Mauna Loa CO2 regression | test MSE | No clear effect |
| Add Matérn 5/2 to Matérn 3/2 (ablation) | Single-output GP · Mauna Loa CO2 regression | test MSE | No clear effect |
| Add the less smooth Matérn 1/2 component to Matérn 3/2 (ablation) | Single-output GP · Mauna Loa CO2 regression | prediction performance | Hurt |
| Add the smoother Matérn 5/2 component to Matérn 3/2 (ablation) | Single-output GP · Mauna Loa CO2 regression | prediction performance | No clear effect |
| Use the additive mixture rather than the least smooth component alone with limited training data (ablation) | Single-output GP · Mauna Loa CO2 regression, training sample size below 50% | test MSE | Mixed |
| Allow Matérn mixture components to have distinct smoothness | Single-output GP · Asymptotic fixed-domain setting, p=1,2,3 | parameter identifiability | Hurt |
| Use Matérn mixture components with the same smoothness (ablation) | Single-output GP · Asymptotic fixed-domain setting | parameter identifiability | Hurt |
| Add Gaussian nugget noise with the same variance across compared models (ablation) | Single-output GP with additive Matérn mixture · Identifiability | parameter identifiability | No clear effect |
| Use a separable multiplicative kernel with covariance matrix A | Multi-output GP · Matérn base kernel | correlation-structure identifiability | Helped |
Closest papers by meaning
- Revisiting Nonstationary Kernel Design for Multi-Output Gaussian Processes (ICLR 2026 · similarity 0.90)
- Scalable Log Determinants for Gaussian Process Kernel Learning (NeurIPS 2017 · similarity 0.89)
- Additive Gaussian Processes Revisited (ICML 2022 · similarity 0.89)
- Solving High Frequency and Multi-Scale PDEs with Gaussian Processes (ICLR 2024 · similarity 0.86)
- On the Consistency of Kernel Methods with Dependent Observations (ICML 2024 · similarity 0.86)
- The Price of Linear Time: Error Analysis of Structured Kernel Interpolation (ICML 2025 · similarity 0.86)
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.