Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Test-time Collective Prediction
Celestine Mendler-Dünner, Wenshuo Guo, Stephen Bates, Michael Jordan
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 DeGroot adaptive weighting instead of equal-weight model averaging | ensemble of linear models · Synthetic logistic | MSE: over 50x reduction | Improved |
| use DeGroot consensus instead of picking best individual model | ensemble · Synthetic logistic | MSE: 5e^{-2} -> 4.6e^{-4} | Improved |
| replace iterative power pooling with averaging trust scores (tau-avg) (ablation) | DeGroot · Synthetic logistic | MSE: 4.6e^{-4} -> 9e^{-3} | Hurt |
| replace DeGroot weights with inverse of dataset-averaged MSE (ablation) | DeGroot · Synthetic logistic | MSE: 4.6e^{-4} -> 1e^{-2} | Hurt |
| increase covariance of local feature distributions (more overlap) (ablation) | DeGroot vs M-avg · Synthetic logistic | gain over M-avg | Hurt |
| make local data variance extremely small (ablation) | DeGroot · Synthetic logistic | gain over M-avg | Hurt |
| choose neighbor count N too large (ablation) | DeGroot · Synthetic logistic | MSE | Hurt |
| choose very small N under highly noisy labels (ablation) | DeGroot · Synthetic logistic | MSE | Hurt |
| vary N between 1% and 10% of local data (ablation) | DeGroot · Synthetic logistic | MSE | No clear effect |
| increase label/feature/model heterogeneity across agents (ablation) | DeGroot vs M-avg · Abalone lasso | relative gain over M-avg | Improved |
| DeGroot vs M-avg on benchmarks (no validation data) | DeGroot · Boston Ridge | gain relative to DeGroot [%]: -12.45 | Improved |
| DeGroot vs CV-adaptive that uses extra validation data | DeGroot · E2006, Abalone, cpusmall, YearPrediction | gain relative to DeGroot [%]: -0.39 +/- 0.75 | No clear effect |
Closest papers by meaning
- Scalable Decentralized Algorithms for Online Personalized Mean Estimation (AAAI 2025 · similarity 0.87)
- A Statistical Framework for Algorithmic Collective Action with Multiple Collectives (NeurIPS 2026 · similarity 0.85)
- Collaborative Mean Estimation Among Heterogeneous Strategic Agents: Individual Rationality, Fairness, and Truthful Contribution (ICML 2025 · similarity 0.85)
- The Role of Learning Algorithms in Collective Action (ICML 2024 · similarity 0.85)
- Stochastically Dominant Peer Prediction (NeurIPS 2025 · similarity 0.84)
- Statistical Collusion by Collectives on Learning Platforms (ICML 2025 · 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.
