Research map / Learning theory and algorithms / Fairness
Is There a Trade-Off Between Fairness and Accuracy? A Perspective Using Mismatched Hypothesis Testing
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 |
|---|---|---|---|
| Enforce equal opportunity on the observed data by adjusting the unprivileged group's threshold | Likelihood ratio classifier · Numerical example, Z=0 observed data | Chernoff exponent of overall error: 9/8 -> 1/2 | Hurt |
| Evaluate the fair detector against ideal distributions instead of the observed distributions | Fair likelihood ratio classifier · Numerical example, Z=0 ideal distributions | Chernoff exponent of overall error: 1/2 -> 2 | Helped |
| Collect an additional feature informative about the label given X | Unprivileged-group classifier · Numerical example, Z=0 | Chernoff information: 9/8 -> 5/4 | Helped |
| Collect an additional feature before enforcing equal opportunity | Fair likelihood ratio classifier · Numerical example, Z=0 | Chernoff exponent of overall error: 0.5 -> 0.68 | Helped |
| Use a fairness constraint when groups have unequal observed separability | Bayes optimal group classifiers · Observed distributions | Equal-opportunity fairness | Hurt |
| Collect additional features with I(X';Y|X,Z=0)>0 | Unprivileged-group classifier · General active data collection setting | Chernoff information | Helped |
Closest papers by meaning
- Unlocking Fairness: a Trade-off Revisited (NeurIPS 2019 · similarity 0.91)
- Achievable Fairness on Your Data With Utility Guarantees (NeurIPS 2024 · similarity 0.90)
- How Far Can Fairness Constraints Help Recover From Biased Data? (ICML 2024 · similarity 0.88)
- Accuracy vs. Accuracy: Computational Tradeoffs Between Classification Rates and Utility (NeurIPS 2026 · similarity 0.87)
- Mitigating Label Bias in Machine Learning: Fairness through Confident Learning (AAAI 2024 · similarity 0.87)
- Reconciling Predictive and Statistical Parity: A Causal Approach (AAAI 2024 · similarity 0.87)
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.