Research map / Learning theory and algorithms / Prediction and decision losses
Joint Manifold Diffusion for Combining Predictions on Decoupled Observations
Kwang In Kim, Hyung Jin Chang
Check 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 |
|---|---|---|---|
| Combine the main predictor with reference predictors using joint manifold diffusion | Relative attribute ranking · AWA | ranking accuracy | Improved |
| Combine predictions from decoupled feature observations | Joint manifold diffusion · Cal7, NUS, and HW | ranking accuracy | Improved |
| Use joint manifold diffusion instead of TPC | Relative attribute ranking · OSR | ranking accuracy | No clear effect |
| Use joint manifold diffusion instead of TPC | Relative attribute ranking · PubFig | average ranking accuracy | Improved |
| Use joint manifold diffusion instead of TPC | Relative attribute ranking · PubFig | ranking accuracy | Mixed |
| Use joint manifold diffusion instead of Ind | Relative attribute ranking · OSR and PubFig | ranking accuracy | Improved |
| Use joint manifold diffusion instead of MTL¹ and MTL² | Relative attribute ranking · PubFig and Shoes | ranking accuracy | Improved |
| Use MTL¹ or MTL² instead of Ind | Test-time predictor combination · Relative attribute ranking | ranking accuracy | Improved |
| Use MTL¹ or MTL² instead of joint manifold diffusion | Relative attribute ranking · OSR attributes 2, 4, and 5 | average ranking accuracy | Mixed |
| Apply bridge estimation and joint diffusion to decoupled synthetic observations | Joint manifold diffusion · Synthetic task-dependence dataset | task-dependence recovery | Improved |
| Replace TPC with the Eigen-decomposition-based joint diffusion computation | AWA predictor combination · AWA | runtime speedup: around 20× | Improved |
Closest papers by meaning
- Predictor Combination at Test Time (ICCV 2017 · similarity 0.90)
- Improving Predictors via Combination Across Diverse Task Categories (ICML 2021 · similarity 0.89)
- Combining Task Predictors via Enhancing Joint Predictability (ECCV 2020 · similarity 0.88)
- Diffusion Models and the Manifold Hypothesis: Log-Domain Smoothing is Geometry Adaptive (NeurIPS 2025 · similarity 0.85)
- Efficient Pareto Manifold Learning with Low-Rank Structure (ICML 2024 · similarity 0.84)
- Analysing Multi-Task Regression via Random Matrix Theory with Application to Time Series Forecasting (NeurIPS 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.
