Research map / Learning theory and algorithms / Fairness
ProbLog4Fairness: A Neurosymbolic Approach to Modeling and Mitigating Bias
Rik Adriaensen, Lucas Van Praet, Jessa Bekker, Robin Manhaeve, Pieter Delobelle, Maarten Buyl
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 unbiased cleaned labels instead of original labels for fine-tuning | ResNet-50 · CelebA Mouth Slightly Open | F1 on cleaned labels: +10.18 % | Improved |
| add ProbLog label-bias program during training | neural classifier · synthetic data, label bias | accuracy and statistical disparity | Improved |
| add ProbLog program that models only the problematic bias when A is not independent of Y | neural classifier · synthetic data, A not independent of Y | statistical disparity | Improved |
| set program bias parameter to the true bias probability rather than a wrong one (ablation) | ProbLog4Fairness · synthetic data, label bias 0.3 | accuracy and statistical disparity | Improved |
| add ProbLog label-bias modeling instead of baselines | neural classifier · Student Alcohol Consumption | unbiased-label F1 | Improved |
| simplify the program to no positive bias or no bias on A=0 (ablation) | ProbLog4Fairness · Student Alcohol Consumption | F1 and statistical disparity | Improved |
| correct for three correlated attributes jointly instead of a single one (ablation) | ProbLog4Fairness · CelebA Mouth Slightly Open | F1 and statistical disparity | Improved |
| add ProbLog bias modeling with single sensitive attribute | ProbLog4Fairness · CelebA Mouth Slightly Open | F1 | Mixed |
Closest papers by meaning
- Prompting Fairness: Integrating Causality to Debias Large Language Models (ICLR 2025 · similarity 0.87)
- Biases Mitigation and Expressiveness Preservation in Language Models: A Comprehensive Pipeline (Student Abstract) (AAAI 2024 · similarity 0.87)
- The Devil is in the Neurons: Interpreting and Mitigating Social Biases in Language Models (ICLR 2024 · similarity 0.87)
- Intersectional Unfairness Discovery (ICML 2024 · similarity 0.86)
- Identifying Reasons for Bias: An Argumentation-Based Approach (AAAI 2024 · similarity 0.85)
- Causal Bias Detection in Generative Artifical Intelligence (NeurIPS 2026 · 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.
