Mitigating Algorithmic Bias on Facial Expression Recognition
This addresses fairness issues for minority groups in facial recognition, but appears incremental as it applies an existing method to a specific domain.
The paper tackles the problem of algorithmic bias in facial expression recognition by proposing a debiasing variational autoencoder to mitigate bias from biased datasets, achieving unspecified results without concrete numbers.
Biased datasets are ubiquitous and present a challenge for machine learning. For a number of categories on a dataset that are equally important but some are sparse and others are common, the learning algorithms will favor the ones with more presence. The problem of biased datasets is especially sensitive when dealing with minority people groups. How can we, from biased data, generate algorithms that treat every person equally? This work explores one way to mitigate bias using a debiasing variational autoencoder with experiments on facial expression recognition.