Research map / Label-efficient and robust learning / Domain adaptation and generalization
Regularized Multi-Output Gaussian Convolution Process With Domain Adaptation
Xinming Wang, Chao Wang, Xuan Song, Levi Kirby, Jianguo Wu
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 |
|---|---|---|---|
| add L1 regularization on source-target kernel scales (MGCP-R vs MGCP) (ablation) | MGCP · Sim III setting 1 (41 outputs, n_e=10) | MAE: 1.211 -> 0.273 | Improved |
| add L1 regularization on source-target kernel scales (ablation) | MGCP · Sim III setting 1 (17 outputs, n_e=4) | MAE: 0.865 -> 0.323 | Improved |
| add L1 regularization on source-target kernel scales (ablation) | MGCP · Sim III setting 1 (9 outputs, n_e=2) | MAE: 0.543 -> 0.251 | Improved |
| add L1 regularization on source-target kernel scales (ablation) | MGCP · Sim III setting 2 (7 outputs, n_e=2) | MAE: 0.556 -> 0.377 | Improved |
| add L1 regularization on source-target kernel scales (ablation) | MGCP · Sim III setting 2 (4 outputs, n_e=1) | MAE: 0.575 -> 0.349 | Improved |
| use unregularized multi-source transfer (MGCP) instead of target-only GCP | GCP · Sim III setting 1 (41 outputs, n_e=10) | MAE: 0.652 -> 1.211 | Hurt |
| add a full covariance among sources (MGCP-RF instead of MGCP-R) (ablation) | MGCP-R · Sim III setting 1 (41 outputs, n_e=10) | MAE: 0.273 -> 0.667 | Hurt |
| add a full covariance among sources (MGCP-RF instead of MGCP-R) (ablation) | MGCP-R · Sim III setting 1 (41 outputs, n_e=10) | optimization time: 62.90 -> 3673.4 | Hurt |
| add regularization to MGCP (ablation) | MGCP · Sim III setting 1 (41 outputs, n_e=10) | optimization time: 281.29 -> 62.90 | Improved |
| use pairwise two-stage integration (BGCP-R) instead of global regularization | MGCP-R · Sim III setting 1 (41 outputs, n_e=10) | MAE: 0.273 -> 0.495 | Hurt |
| use pairwise two-stage BGCP-R instead of target-only GCP | GCP · Sim I (1D example) | MAE | Hurt |
| add L1 penalty on scale of uninformative source kernel (ablation) | MGCP · Sim I 1D example, alpha_3t | scale parameter: 2.53 -> 1.00e-5 | Improved |
Closest papers by meaning
- Minimizing Negative Transfer of Knowledge in Multivariate Gaussian Processes: A Scalable and Regularized Approach (TPAMI 2020 · similarity 0.90)
- Unsupervised Visual Domain Adaptation: A Deep Max-Margin Gaussian Process Approach (CVPR 2019 · similarity 0.88)
- Adaptive Transfer Kernel Learning for Transfer Gaussian Process Regression (TPAMI 2022 · similarity 0.86)
- A Bayesian Hierarchical Framework for Non-Linear and Iterative Transfer Learning of Gaussian Process (TPAMI 2026 · similarity 0.86)
- Leveraging Normalization Layer in Adapters with Progressive Learning and Adaptive Distillation for Cross-Domain Few-Shot Learning (AAAI 2024 · similarity 0.83)
- Bayesian Domain Adaptation with Gaussian Mixture Domain-Indexing (NeurIPS 2024 · 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.
