Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Longitudinal Deep Kernel Gaussian Process Regression
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 a time-invariant component for individual-level effects (ablation) | L-DKGPR · TADPOLE, SWAN, and GSS full data | R² (%): 64.9±1.4 vs 13.2±1.1; 52.5±0.4 vs 29.0±3.2; 56.9±0.1 vs 56.2±0.1 | Helped |
| Add a time-varying component (ablation) | L-DKGPR · TADPOLE, SWAN, and GSS full data | R² (%): 64.9±1.4 vs 56.3±1.3; 52.5±0.4 vs 16.7± 2.4; 56.9±0.1 vs -0.2±0.2 | Helped |
| Use a learned deep kernel instead of an RBF kernel (ablation) | L-DKGPR · TADPOLE, SWAN, and GSS full data | R² (%): 64.9±1.4 vs 55.5±2.4; 52.5±0.4 vs 5.4±1.6; 56.9±0.1 vs -14.1±0.4 | Helped |
| Use the closed-form Algorithm solver rather than Monte Carlo sampling with M=10 and 300 iterations (ablation) | L-DKGPR · SWAN and GSS full data | R² (%): 52.5±0.4 vs 3.1±0.2; 56.9±0.1 vs 4.5±0.1 | Helped |
| Use Monte Carlo sampling with M=128 and 3,000 iterations instead of the closed-form solver (ablation) | L-DKGPR · SWAN and GSS full data | R² (%): 52.5±0.4 vs 51.4±0.4; 56.9±0.1 vs 55.6±0.1 | Hurt |
| Use L-DKGPR rather than the strongest reported baseline | Longitudinal regression · Simulated MC, C=2 through C=5 | R² (%): 91.3±0.2 vs -1.6±16.9; 99.6±0.2 vs -14.7±6.5; 99.8±0.2 vs -13.5±8.4; 99.8±0.2 vs -6.1±4.4 | Helped |
| Use L-DKGPR rather than the strongest reported baseline | Longitudinal regression · TADPOLE, full data | R² (%): 64.9±1.4 vs 61.9±1.9 | Helped |
| Use L-DKGPR rather than the strongest reported baseline | Longitudinal regression · SWAN and GSS, full data | R² (%): 52.5±0.4 vs 48.6±2.0; 56.9±0.1 vs 54.8±2.2 | Helped |
| Use L-DKGPR rather than GLMM | Longitudinal regression · TADPOLE, 50 individuals | R² (%): 44.0±5.6 vs 50.8±5.5 | Hurt |
| Optimize inducing points jointly in latent space | L-DKGPR · Simulated data | R² (%) | Helped |
Closest papers by meaning
- Inducing Clusters Deep Kernel Gaussian Process for Longitudinal Data (AAAI 2024 · similarity 0.91)
- LMLFM: Longitudinal Multi-Level Factorization Machine (AAAI 2020 · similarity 0.86)
- Adaptive Shrinkage Estimation for Personalized Deep Kernel Regression in Modeling Brain Trajectories (ICLR 2025 · similarity 0.85)
- KernelMatmul: Scaling Gaussian Processes to Large Time Series (AAAI 2025 · similarity 0.85)
- Kernel Representation Learning for Time Sequence: Algorithm, Theory, and Applications (AAAI 2025 · similarity 0.84)
- Latent variable model for high-dimensional point process with structured missingness (ICML 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.