PLS-based approach for fair representation learning
This work addresses fairness in machine learning for applications requiring unbiased predictions, but it is incremental as it builds on existing PLS and fairness techniques.
The authors tackled the problem of fair representation learning by proposing Fair Partial Least Squares (PLS) components, which incorporate fairness constraints into PLS to reduce data dimensionality for prediction, and demonstrated its superiority over standard fair PCA methods on various datasets.
We revisit the problem of fair representation learning by proposing Fair Partial Least Squares (PLS) components. PLS is widely used in statistics to efficiently reduce the dimension of the data by providing representation tailored for the prediction. We propose a novel method to incorporate fairness constraints in the construction of PLS components. This new algorithm provides a feasible way to construct such features both in the linear and the non linear case using kernel embeddings. The efficiency of our method is evaluated on different datasets, and we prove its superiority with respect to standard fair PCA method.