MSc Thesis Proposal - “vivaFemme: Mitigating Gender Bias in Neural Team Formation via Female-Advocate Loss Regularization” by: Roonak Moasses

Friday, March 8, 2024 - 15:00 to 16:30

vivaFemme: Mitigating Gender Bias in Neural Team Formation via Female-Advocate Loss Regularization”

MSc Thesis Proposal by: Roonak Moasses

 

Date: Friday, 08 Mar 2024

Time:  3:00 pm - 4:30 pm

Location: Essex Hall, Room 122

 

Abstract:
Neural team formation has brought state-of-the-art efficacy while enhancing efficiency at forming teams of experts whose success in completing complex tasks is almost surely guaranteed, yet proposed methods overlook diversity; that is, predicted teams are male dominant and females’ participation is scarce. To this end, pre- and post-processing debiasing techniques have been initially proposed, mainly for being model-agnostic with little to no modification to the model’s architecture. Though their poor mitigation performance have proven futile especially in the presence of extreme bias, e.g., 95\% male vs. 5\% female experts in the training datasets, urging further development of in-process debiasing techniques. In this proposal, the first of its kind, we aim to develop an in-process gender debiasing method in neural team formation via modifications to models’ conventional cross-entropy loss function. Specifically, (1) we penalize the model (ie., an increase to the loss) for false negative female experts, and meanwhile, (2) we randomly sample from female experts and reinforce the likelihood of female participation in the predicted teams, even at the cost of increasing false positive female. Our early experiment on a large-scale benchmark dataset (IMDB) withholding extreme gender bias shows our method’s competence in mitigating neural models’ gender bias while maintaining accuracy, resulting in diverse yet successful teams. The code to reproduce the experiments is available at https://github.com/fani-lab/OpeNTF/tree/VivaFemme.
Keywords: Team Formation, Fairness, Neural Networks
 
Thesis Committee:
Internal Reader: Dr. Saeed Samet, School of Computer Science
External Reader: Dr. Charlene Y. Senn, Canada Research Chair, Dept. of Psychology/Women’s and Gender Studies
Advisor: Dr. Hossein Fani, School of Computer Science
 
Vector Institute Logo