Roadmap for gender bias mitigation in OpeNTF

[Rounique]

Steps and tasks to be done for gender bias mitigation in OpeNTF:

1. Related works

• Learning to form skill-based teams of experts.
• Bootless Application of Greedy Re-ranking Algorithms in Fair Neural Team Formation
• Increasing Fairness in Predictions Using Bias Parity Score-Based Loss Function Regularization
• Deep neural networks for optimal team composition
• Weight uncertainty in neural network
  .
  .
  .

2. Experiments

• Increasing the female ratio in training sets manually #230
• Adjusting the loss function #231
• Adjusting the loss function + adding the NDKL fairness metric #232
• Experiment grid on weights and ndkl's ratios on imdb #233
• Experimenting with very high weights #235
• [ ]
  .
  .
  .

4. Write up

• Abstract
• Introduction
• Related works
• Problem definition
• Methodology
• Results
  .
  .
  .

[Rounique]

Hi @hosseinfani, here are my previous works and latest updates.