Research map / Label-efficient and robust learning / Domain adaptation and generalization
Adaptive Transfer Kernel Learning for Transfer Gaussian Process Regression
Pengfei Wei, Yiping Ke, Yew-Soon Ong, Zejun Ma
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 |
|---|---|---|---|
| use multiple primitive kernels with separate coefficients (TrGPαβ) instead of single coefficient (TrGPλ) | TrGP · Synthetic (0,0.8) | RMSE: 0.293 -> 0.171 | Improved |
| use multiple primitive kernels with separate coefficients (TrGPαβ) instead of single coefficient (TrGPλ) | TrGP · Synthetic (0.8,1) | RMSE: 0.083 -> 0.037 | Improved |
| use TrGPαβ instead of TrGPλ when sim1 = sim2 | TrGP · Synthetic (0.8,0.8) | RMSE: 0.061 -> 0.060 | No clear effect |
| use TrGPαβ instead of TrGPλ when sim1 = sim2 | TrGP · Synthetic (1,1) | RMSE: 0.001 -> 0.001 | No clear effect |
| transfer from source with TrGPαβ versus target-only GP | GPtar · Synthetic (0,0) | RMSE: 0.296 -> 0.279 | No clear effect |
| transfer from highly related source with TrGPαβ versus target-only GP | GPtar · Synthetic (1,1) | RMSE: 0.296 -> 0.001 | Improved |
| learned λ is a compromise when sim1 differs from sim2 | TrGPλ · Synthetic (0,0.8) | RMSE: 0.293 | Hurt |
| use polynomial/neural transfer kernel TrGPω instead of linear TrGPαβ | TrGPαβ · Airline time series extrapolation | extrapolation quality | Improved |
| combine source and target data naively (GPall) instead of transfer kernel | GPtar · Airline time series extrapolation | extrapolation quality | Hurt |
| use TrGPω instead of TrGPλ, TrGPλvar and other baselines | TrGPαβ · Wine, Energy, Air Quality, UJIIndoorLoc | RMSE | Improved |
Closest papers by meaning
- Transfer Kernel Learning for Multi-Source Transfer Gaussian Process Regression (TPAMI 2022 · similarity 0.95)
- Source-Target Similarity Modelings for Multi-Source Transfer Gaussian Process Regression (ICML 2017 · similarity 0.88)
- Wasserstein Transfer Learning (NeurIPS 2025 · similarity 0.86)
- Smoothness Adaptive Hypothesis Transfer Learning (ICML 2024 · similarity 0.85)
- On Hypothesis Transfer Learning of Functional Linear Models (ICML 2024 · similarity 0.84)
- Transferring Knowledge From Large Foundation Models to Small Downstream Models (ICML 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.
