Such an important paper, feeling v fomo of icml! “Motivated by geometric considerations, we mitigate gender bias through a new post-processing methodology which transforms the deep embeddings of a pre-trained model to give more representation power to discriminated subgroups.”
Quote Tweet
Want to know more about gender bias in face recognition and how to mitigate them?
Come have a chat with us at the poster session this evening at #ICML2022 !