#FairnessML

DCAST: Diverse Class-Aware Self-Training Mitigates Selection Bias for Fairer Learning Fairness in machine learning seeks to mitigate model bias against individuals based on sensitive features such as sex or age, often caused by an uneven representation of the population in the training...

Paper by our Yasin Tepeli: How to train fairer ML models from data affected by selection bias? Diversity! DCAST learns from diverse vs. most confident samples to avoid confirmation bias via semi-supervised learning. #MachineLearning #FairnessML arxiv.org/abs/2409.20126