Research map / Learning theory and algorithms / Fairness
Fair and Diverse DPP-Based Data Summarization
Elisa Celis, Vijay Keswani, Damian Straszak, Amit Deshpande, Tarun Kathuria, Nisheeth Vishnoi
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 |
|---|---|---|---|
| Enforce equal gender representation with P-DPP | k-DPP on Adult · Adult, Gender, equal representation | D^un (unfairness): 0.027 -> 0 | Improved |
| Enforce equal gender representation with P-DPP | k-DPP on Adult · Adult, Gender, equal representation | log G (diversity): 489 -> 490 | No clear effect |
| Enforce proportional gender representation with P-DPP | k-DPP on Adult · Adult, Gender, proportional representation | D^prop (unfairness): 0.011 -> 0 | Improved |
| Enforce proportional gender representation with P-DPP | k-DPP on Adult · Adult, Gender, proportional representation | log G (diversity): 489 -> 482 | No clear effect |
| Enforce equal race representation with P-DPP | k-DPP on Adult · Adult, Race, equal representation | D^un (unfairness): 0.268 -> 0 | Improved |
| Enforce equal race representation with P-DPP | k-DPP on Adult · Adult, Race, equal representation | log G (diversity): 487 -> 476 | No clear effect |
| Enforce proportional race representation with P-DPP | k-DPP on Adult · Adult, Race, proportional representation | D^prop (unfairness): 0.005 -> 0 | Improved |
| Enforce proportional race representation with P-DPP | k-DPP on Adult · Adult, Race, proportional representation | log G (diversity): 487 -> 488 | No clear effect |
| Scale tail singular values of partition matrices by 1/n (ablation) | k-DPP on a random dataset · Random dataset, before vs after scaling | D^un (unfairness): 0.063 -> 5.2 \times 10^{-6} | Improved |
| Scale tail singular values of partition matrices by 1/n (ablation) | k-DPP on a random dataset · Random dataset, before vs after scaling | log G (diversity): 457.3 -> 397.4 | Hurt |
| Use Scale-And-Sample instead of unconstrained k-DPP | k-DPP on a random dataset · Random dataset, before scaling | D^un (unfairness): 0.063 -> 5.2 \times 10^{-6} | Improved |
| Use P-DPP instead of k_i-DPP | Fair sampling on Adult · Adult, all representation settings | log G (diversity): Gender equal: 410 -> 490; Gender proportional: 409 -> 482; Race equal: 366 -> 476; Race proportional: 426 -> 488 | Improved |
Closest papers by meaning
- FairWASP: Fast and Optimal Fair Wasserstein Pre-processing (AAAI 2024 · similarity 0.84)
- Who’s the (Multi-)Fairest of Them All: Rethinking Interpolation-Based Data Augmentation Through the Lens of Multicalibration (AAAI 2025 · similarity 0.83)
- Fair Wasserstein Coresets (NeurIPS 2024 · similarity 0.83)
- Measures of diversity and space-filling designs for categorical data (ICML 2024 · similarity 0.83)
- Size-adaptive Hypothesis Testing for Fairness (NeurIPS 2025 · similarity 0.82)
- fairret: a Framework for Differentiable Fairness Regularization Terms (ICLR 2024 · similarity 0.82)
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.
