Flexibly Fair Representation Learning by Disentanglement
This addresses fairness in machine learning for applications involving multiple sensitive attributes, but it is incremental as it builds on disentangled representation learning.
The paper tackles the problem of learning representations that achieve group and subgroup fairness with respect to multiple sensitive attributes, proposing an algorithm for flexibly fair representations that can be modified at test time to achieve demographic parity, and shows empirically that the encoder enables adaptation to various fair classification tasks without requiring sensitive attributes for inference.
We consider the problem of learning representations that achieve group and subgroup fairness with respect to multiple sensitive attributes. Taking inspiration from the disentangled representation learning literature, we propose an algorithm for learning compact representations of datasets that are useful for reconstruction and prediction, but are also \emph{flexibly fair}, meaning they can be easily modified at test time to achieve subgroup demographic parity with respect to multiple sensitive attributes and their conjunctions. We show empirically that the resulting encoder---which does not require the sensitive attributes for inference---enables the adaptation of a single representation to a variety of fair classification tasks with new target labels and subgroup definitions.