Research map / Learning theory and algorithms / Fairness
Active Sampling for Min-Max Fairness
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Adaptively sample training points from the currently highest-loss group | Min-max stochastic gradient descent · Drug Consumption and COMPAS | convergence speed | Helped |
| Use accelerated adaptive group reweighting and weighted full-data gradient updates instead of single-point sampling | Min-max optimization · Convex learning problems | convergence rate: ~1/sqrt{T} -> ~1/T | Helped |
| Adaptively sample from the highest-loss group rather than train with standard SGD | MLP on Diabetes · Diabetes test set | worst-group loss | Helped |
| Use a small validation set of 20 datapoints per group instead of a larger validation set (ablation) | Min-max stochastic gradient descent · Drug Consumption | convergence speed | Helped |
| Return the final iterate instead of the average iterate (ablation) | Min-max stochastic gradient descent · Diabetes with MLP | worst-group loss | Helped |
| Blend whole-population sampling with worst-group sampling using probability p (ablation) | Min-max stochastic gradient descent · Drug Consumption | maximum group loss versus population loss | Mixed |
| Use mini-batch updates instead of single-sample stochastic updates (ablation) | Min-max stochastic gradient descent · Convex learning problems | convergence guarantee | No clear effect |
Closest papers by meaning
- Fairness without Harm: An Influence-Guided Active Sampling Approach (NeurIPS 2024 · similarity 0.90)
- Adaptive Sampling for Minimax Fair Classification (NeurIPS 2021 · similarity 0.89)
- Ensuring Fairness Beyond the Training Data (NeurIPS 2020 · similarity 0.87)
- Fairness via Independence: A General Regularization Framework for Machine Learning (ICLR 2026 · similarity 0.87)
- On the Convergence of Loss and Uncertainty-based Active Learning Algorithms (NeurIPS 2024 · similarity 0.86)
- Geometry of Relaxed Fair Regression: A Unified Framework for Aware and Unaware Settings (NeurIPS 2026 · similarity 0.86)
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.