Research map / Learning theory and algorithms / Fairness
Are My Deep Learning Systems Fair? An Empirical Study of Fixed-Seed Training
Shangshu Qian, Hung Viet Pham, Thibaud Lutellier, Zeou Hu, Jungwon Kim, Lin Tan, Yaoliang Yu, Jiahao Chen, Sameena Shah
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 |
|---|---|---|---|
| fix random seed across identical training runs | DL fairness training · 27 techniques, 16 fixed-seed runs | bias metrics (DP, DI, etc.): up to 12.6% DP max difference | No clear effect |
| apply debiasing algorithm GR vs baseline, averaged over 16 runs | S-Base · BA metric | bias amplification (%): 7.4 -> 8.7 | Hurt |
| apply 22 debiasing techniques | baseline DL models · 7 bias metrics, 154 experiments | bias values: 103 of 154 reduce bias; 15 of 22 techniques increase at least one metric | Mixed |
| apply debiasing that reduces bias | baseline DL models · 103 bias-reducing experiments | accuracy and variance costs: 70 of 103 with cost | Hurt |
| apply RBA debiasing (S-DD4) | S-Base · 7 bias metrics | bias and variance | Improved |
Closest papers by meaning
- Benchmarking Bias Mitigation Algorithms in Representation Learning through Fairness Metrics (NeurIPS 2021 · similarity 0.88)
- DeepGemini: Verifying Dependency Fairness for Deep Neural Network (AAAI 2023 · similarity 0.87)
- Achievable Fairness on Your Data With Utility Guarantees (NeurIPS 2024 · similarity 0.85)
- Can Private Machine Learning Be Fair? (AAAI 2025 · similarity 0.85)
- Adversarial Fairness Network (AAAI 2024 · similarity 0.85)
- On the Alignment between Fairness and Accuracy: from the Perspective of Adversarial Robustness (ICML 2025 · 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.
