Research map / Learning theory and algorithms / Fairness
DeepGemini: Verifying Dependency Fairness for Deep Neural Network
Xuan Xie, Fuyuan Zhang, Xinwen Hu, Lei Ma
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 |
|---|---|---|---|
| Use DeepGemini's SM-random fairness analysis instead of Libra-Symbolic | COMPAS DNNs with five hidden layers · COMPAS | fairness score (%): 51.48% vs 26.79% | Improved |
| Apply iterative region refinement in CEGFA | DeepGemini · German credit and COMPAS | fairness score | Improved |
| Train on the fair rather than biased dataset version | DNNs analyzed by DeepGemini · German credit and COMPAS | fairness score | Improved |
| Use random, saliency-map, important-neuron, or ε-greedy region refinement (ablation) | CEGFA · German credit and COMPAS | fairness score | Mixed |
| Increase DNN size from four hidden layers with 20 neurons per layer to four hidden layers with 30 neurons per layer | DeepGemini · COMPAS; query QC1 | verification time (seconds): 37.0 -> 225.96 | Hurt |
Closest papers by meaning
- FAIRER: Fairness as Decision Rationale Alignment (ICML 2023 · similarity 0.90)
- Towards Debiasing DNN Models from Spurious Feature Influence (AAAI 2022 · similarity 0.88)
- Adversarial Fairness Network (AAAI 2024 · similarity 0.86)
- Fair Bilevel Neural Network (FairBiNN): On Balancing fairness and accuracy via Stackelberg Equilibrium (NeurIPS 2024 · similarity 0.86)
- Some Optimizers are More Equal: Understanding the Role of Optimizers in Group Fairness (NeurIPS 2025 · similarity 0.86)
- FairNet: Dynamic Fairness Correction without Performance Loss via Contrastive Conditional LoRA (NeurIPS 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.
