Group items tagged

The blind application of machine learning runs the risk of amplifying biases present in data. Such a danger is facing us with word embedding, a popular framework to represent text data as vectors which has been used in many machine learning and natural language processing tasks. We show that even word embeddings trained on Google News articles exhibit female/male gender stereotypes to a disturbing extent. This raises concerns because their widespread use, as we describe, often tends to amplify these biases. Geometrically, gender bias is first shown to be captured by a direction in the word embedding. Second, gender neutral words are shown to be linearly separable from gender definition words in the word embedding. Using these properties, we provide a methodology for modifying an embedding to remove gender stereotypes, such as the association between between the words receptionist and female, while maintaining desired associations such as between the words queen and female. We define metrics to quantify both direct and indirect gender biases in embeddings, and develop algorithms to "debias" the embedding. Using crowd-worker evaluation as well as standard benchmarks, we empirically demonstrate that our algorithms significantly reduce gender bias in embeddings while preserving the its useful properties such as the ability to cluster related concepts and to solve analogy tasks. The resulting embeddings can be used in applications without amplifying gender bias.

If you searched for "three white teenagers" on Google Images earlier this month, the result spat up shiny, happy people in droves - an R.E.M. song in JPG format. The images, mostly stock photos, displayed young Caucasian men and women laughing, holding sports equipment or caught whimsically mid-selfie. If you searched for "three black teenagers," the algorithm offered an array of mug shots.

Search term brings back mainly results of black women, which some say is evidence of bias. But algorithms may just be reflecting the wider social landscape

Algorithms used in American criminal justice is found to be biased against blacks

Google came under fire this week after its new Photos app categorized photos in one of the most racist ways possible. On June 28th, computer programmer Jacky Alciné found that the feature kept tagging pictures of him and his girlfriend as "gorillas."