2017

Data Decisions and Theoretical Implications when Adversarially Learning Fair Representations

Beutel, Alex, Chen, Jilin, Zhao, Zhe et al.

Understand

How can we learn a classifier that is "fair" for a protected or sensitive group, when we do not know if the input to the classifier belongs to the protected group? How can we train such a classifier when data on the protected group is difficult to attain? In many settings, finding out the sensitive input attribute can be prohibitively expensive even during model training, and sometimes impossible during model serving.

  • For example, in recommender systems, if we want to predict if a user will click on a given recommendation, we often do not know many attributes of the user, e.g., race or age, and many attributes of the content are hard to determine, e.g., the language or topic.
  • Thus, it is not feasible to use a different classifier calibrated based on knowledge of the sensitive attribute.
  • Here, we use an adversarial training procedure to remove information about the sensitive attribute from the latent representation learned by a neural network.

Built on

  • Adaptive subgradient methods for online learning and stochastic optimization

    John Duchi, Elad Hazan, and Yoram Singer. 2011 · 2011

    Earlier work this paper cites.

  • UCI Machine Learning Repository

    M. Lichman. 2013 · 2013

    Earlier work this paper cites.

  • Learning fair representations. In Proceedings of the 30th International Conference on Machine Learning (ICML-13)

    Rich Zemel, Yu Wu, Kevin Swersky, Toni Pitassi, and Cynthia Dwork. 2013 · 2013

    Earlier work this paper cites.

  • Censoring representations with an adversary

    Original

    Harrison Edwards and Amos Storkey. 2015 · 2015

    Earlier work this paper cites.

Similar

  • The variational fair autoencoder

    Original

    Christos Louizos, Kevin Swersky, Yujia Li, Max Welling, and Richard Zemel. 2015 · 2015

    Cited alongside, same era.

  • Man is to computer programmer as woman is to homemaker? debiasing word embeddings. In Advances in Neural Information Processing Systems

    Tolga Bolukbasi, Kai-Wei Chang, James Y Zou, Venkatesh Saligrama, and Adam T Kalai. 2016 · 2016

    Cited alongside, same era.

  • Domain separation networks. In Advances in Neural Information Processing Systems

    Konstantinos Bousmalis, George Trigeorgis, Nathan Silberman, Dilip Krishnan, and Dumitru Erhan. 2016 · 2016

    Cited alongside, same era.

  • Domain-adversarial training of neural networks

    Yaroslav Ganin, Evgeniya Ustinova, Hana Ajakan, Pascal Germain, Hugo Larochelle, François Laviolette, Mario Marchand, and Victor Lempitsky. 2016 · 2016

    Cited alongside, same era.

Then

  • Equality of opportunity in supervised learning. In Advances in Neural Information Processing Systems

    Moritz Hardt, Eric Price, Nati Srebro, et al · 2016

    Later among the works it cites.

  • Inherent trade-offs in the fair determination of risk scores

    Original

    Jon Kleinberg, Sendhil Mullainathan, and Manish Raghavan. 2016 · 2016

    Later among the works it cites.

  • Beyond Globally Optimal: Focused Learning for Improved Recommendations. In Proceedings of the 26th International Conference on World Wide Web

    Alex Beutel, Ed H Chi, Zhiyuan Cheng, Hubert Pham, and John Anderson. 2017 · 2017

    Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…