Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Active Assessment of Prediction Services as Accuracy Surface Over Attribute Combinations
Vihari Piratla, Soumen Chakrabarti, Sunita Sarawagi
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 Dirichlet supervision of Beta scale to BetaGP (ablation) | BetaGP · AC-COCOS10K, AC-COCOS, MF-IMDB, MF-CelebA | macro MSE: 7.1 / 0.3 -> 5.3 / 0.2; 3.3 / 0.2 -> 2.8 / 0.0; 2.2 / 0.2 -> 1.4 / 0.1; 4.6 / 0.8 -> 4.1 / 0.1 | Improved |
| Add sparse-arm observation pooling to BetaGP-SL (ablation) | BetaGP-SL · AC-COCOS10K, AC-COCOS, MF-IMDB, MF-CelebA | worst MSE: 11.7 / 1.7 -> 10.4 / 1.5; 6.8 / 0.5 -> 5.7 / 0.5; 22.6 / 1.4 -> 23.3 / 2.3; 4.4 / 0.6 -> 3.9 / 0.7 | Mixed |
| Use Beta mean and scale parameters rather than standard (alpha, beta) parameters (ablation) | BetaGP · MF-CelebA and AC-COCOS | macro MSE: 6.2 / 0.1 -> 4.6 / 0.8; 4.8 / 0.1 -> 3.3 / 0.2 | Improved |
| Smooth accuracy across related arms using a GP (ablation) | Beta-I · MF-CelebA | worst MSE: 8.4 / 0.7 -> 7.6 / 1.3 | Improved |
| Calibrate attribute probabilities with temperature scaling and a joint potential (ablation) | Raw attribute predictions · Four classification tasks | macro MSE | Improved |
| Add temperature calibration without the joint-potential term (ablation) | Raw attribute predictions · Four classification tasks | macro MSE | Improved |
| Use GP-smoothed mean and scale to model arm-specific uncertainty instead of Bernoulli GP classification (ablation) | BernGP · Four classification tasks | estimation quality | Improved |
| Use BetaGP for variance-based exploration without scale supervision or pooling (ablation) | BetaGP · Four classification tasks | macro MSE during exploration | Hurt |
Closest papers by meaning
- Boosting Test Performance with Importance Sampling--a Subpopulation Perspective (AAAI 2025 · similarity 0.82)
- Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensembles (NeurIPS 2021 · similarity 0.82)
- Prediction-Powered Active Testing (NeurIPS 2026 · similarity 0.81)
- SureMap: Simultaneous mean estimation for single-task and multi-task disaggregated evaluation (NeurIPS 2024 · similarity 0.81)
- Computation-Aware Gaussian Processes: Model Selection And Linear-Time Inference (NeurIPS 2024 · similarity 0.81)
- Minimal Variance Model Aggregation: A principled, non-intrusive, and versatile integration of black box models (ICLR 2025 · similarity 0.81)
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.
