Research map / Learning theory and algorithms / Fairness
Incentivizing Truthfulness and Collaborative Fairness in Bayesian Learning
Rachael Hwee Ling Sim, Jue Fan, Xiao Tian, Xinyi Xu, Patrick Jaillet, Bryan Kian Hsiang Low
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 |
|---|---|---|---|
| submit only 50% subset, noisy outputs, 3x duplicated, injected mislabeled data or noisy inputs instead of the true dataset | validation log-likelihood DVF · GP-FR, LO-HE, NN-CY (3 sources) | value of source 0 | Hurt |
| increase noise added to validation set outputs (misspecified validation likelihood) (ablation) | validation log-likelihood DVF · GP-FR, LO-HE, NN-CY | truthfulness of v(D_0) | Hurt |
| use larger validation sets (ablation) | validation log-likelihood DVF · GP-FR, LO-HE, NN-CY | variance of data value | Improved |
| source 0 submits untruthful data (subset or noise) instead of true data | Shapley value semivalue reward · GP-FR, LO-HE, NN-CY | Shapley values of other sources | Improved |
| put larger weight on smaller coalitions in the semivalue (ablation) | semivalue reward · appendix semivalue weights | collaborative fairness | Hurt |
| cap reward at budget B=.06 (limited budget) (ablation) | semivalue reward · GP-FR | truthfulness: B = .06 | Mixed |
| remove validation set by partitioning each source's data 75%-25% into train and validation (ablation) | semivalue reward · GP-FR, LO-HE, NN-CY | reward of source 0: 75%-25% split | No clear effect |
| source 0 injects mislabeled data (strategy I) in the validation-set-free setting (ablation) | semivalue reward without validation set · GP-FR, LO-HE, NN-CY | rewards of sources 1 and 2 | Hurt |
Closest papers by meaning
- Incentivizing Time-Aware Fairness in Data Sharing (NeurIPS 2025 · similarity 0.88)
- A Cramér–von Mises Approach to Incentivizing Truthful Data Sharing (NeurIPS 2025 · similarity 0.88)
- Carrot and Stick: Eliciting Comparison Data and Beyond (NeurIPS 2024 · similarity 0.85)
- Fairness without Harm: An Influence-Guided Active Sampling Approach (NeurIPS 2024 · similarity 0.85)
- You Get What You Give: Reciprocally Fair Federated Learning (ICML 2025 · similarity 0.85)
- Achievable Fairness on Your Data With Utility Guarantees (NeurIPS 2024 · similarity 0.85)
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.
