Diversity in Data Sets
We are continuing to make our Lenses work better for all members of our community. That starts with using diverse training and evaluation datasets in our machine learning (ML) products that power our augmented reality experiences.
We want our ML to be more inclusive, and are educating our team members and providing them with tools to build and maintain more inclusive data sets. The initial focus of these efforts is on the intersection between skin tone, age, and gender. Our plan is to eventually expand this program to all our new and existing ML-based products.