Data Literacy for Responsible AI: Algorithmic Bias
DataRobot
AUGUST 16, 2021
Mitigation strategies can take place at three different stages of the machine learning pipeline: Pre-processing: aimed at reducing bias in the data used to train a model In-processing: applied during training, forcing models to learn less biased patterns Post-processing: modifying model predictions based on fairness metrics.
Let's personalize your content