Research map / Label-efficient and robust learning / Domain adaptation and generalization
A Probabilistic Approach to Knowledge Translation
Shangpu Jiang, Daniel Lowd, Dejing Dou
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 heuristic structure translation rather than learning structure from knowledge samples (ablation) | Knowledge translation on University · University target | WPLL: -12.07 -> -2.51 (1 sample); -3.82 -> -2.80 (2 samples); -3.48 -> -1.79 (5 samples) | Improved |
| Use heuristic structure translation rather than learning structure from knowledge samples (ablation) | Knowledge translation on University · University translated source | WPLL: -9.19 -> -2.05 (1 sample); -3.72 -> -2.10 (2 samples); -1.51 -> -0.97 (5 samples) | Mixed |
| Use manually specified structure with translated source data instead of heuristic structure translation from knowledge | Knowledge translation on University · University target | WPLL: -2.51 -> -1.94 (1 sample); -2.80 -> -1.91 (2 samples); -1.79 -> -1.76 (5 samples) | Mixed |
| Increase translated knowledge samples from 1 to 5 (ablation) | LS-KS · University target | WPLL: -12.07 -> -3.48 | Improved |
| Increase translated knowledge samples from 1 to 5 (ablation) | TS-KS · University target | WPLL: -2.51 -> -1.79 | Improved |
| Use learned target structure rather than an empty structure (ablation) | Knowledge translation on University · University target | WPLL: -3.77 -> -12.07 (1 sample); -3.76 -> -3.82 (2 samples); -3.83 -> -3.48 (5 samples) | Mixed |
| Use knowledge translation rather than translated source data with learned structure | University domain · University target | WPLL: -12.07 vs -3.70 (1 sample); -3.82 vs -3.01 (2 samples) | Mixed |
| Use knowledge translation rather than translated source data with manually specified structure | University domain · University target | WPLL: -2.51 vs -1.94 (1 sample); -2.80 vs -1.91 (2 samples); -1.79 vs -1.76 (5 samples) | Improved |
| Translate knowledge instead of using target data | NBA domain · NBA target data | PLL | Mixed |
| Use learned rather than heuristically translated structure (ablation) | Knowledge translation on NBA · NBA | PLL | No clear effect |
Closest papers by meaning
- Seeking Neural Nuggets: Knowledge Transfer in Large Language Models from a Parametric Perspective (ICLR 2024 · similarity 0.83)
- MergeNet: Knowledge Migration Across Heterogeneous Models, Tasks, and Modalities (AAAI 2025 · similarity 0.83)
- WASD: Wasserstein-based Knowledge Distillation for Large Language Models (NeurIPS 2026 · similarity 0.83)
- Towards Trustworthy Knowledge Graph Reasoning: An Uncertainty Aware Perspective (AAAI 2025 · similarity 0.83)
- BIRD: A Trustworthy Bayesian Inference Framework for Large Language Models (ICLR 2025 · similarity 0.83)
- Knowledge Fusion of Large Language Models (ICLR 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.
