Neutralized Empirical Risk Minimization with Generalization Neutrality Bound
Currently, machine learning plays an important role in the lives and individual activities of numerous people. Accordingly, it has become necessary to design machine learning algorithms to ensure that discrimination, biased views, or unfair treatment do not result from decision making or predictions made via machine learning. In this work, we introduce a novel empirical risk minimization (ERM) framework for supervised learning, neutralized ERM (NERM) that ensures that any classifiers obtained can be guaranteed to be neutral with respect to a viewpoint hypothesis. More specifically, given a viewpoint hypothesis, NERM works to find a target hypothesis that minimizes the empirical risk while simultaneously identifying a target hypothesis that is neutral to the viewpoint hypothesis. Within the NERM framework, we derive a theoretical bound on empirical and generalization neutrality risks. Furthermore, as a realization of NERM with linear classification, we derive a max-margin algorithm, neutral support vector machine (SVM). Experimental results show that our neutral SVM shows improved classification performance in real datasets without sacrificing the neutrality guarantee.
Keywordsneutrality discrimination fairness classification empirical risk minimization support vector machine
Unable to display preview. Download preview PDF.
- 3.Calders, T., Kamiran, F., Pechenizkiy, M.: Building classifiers with independency constraints. In: Saygin, Y., Yu, J.X., Kargupta, H., Wang, W., Ranka, S., Yu, P.S., Wu, X. (eds.) ICDM Workshops, pp. 13–18. IEEE Computer Society (2009)Google Scholar
- 6.Kakade, S.M., Sridharan, K., Tewari, A.: On the complexity of linear prediction: Risk bounds, margin bounds, and regularization. In: Koller, D., Schuurmans, D., Bengio, Y., Bottou, L. (eds.) NIPS, pp. 793–800. Curran Associates, Inc. (2008)Google Scholar
- 7.Kamiran, F., Calders, T., Pechenizkiy, M.: Discrimination aware decision tree learning. In: 2010 IEEE 10th International Conference on Data Mining (ICDM), pp. 869–874. IEEE (2010)Google Scholar
- 10.Pariser, E.: The Filter Bubble: What The Internet Is Hiding From You. Viking, London (2011)Google Scholar
- 11.Pedreschi, D., Ruggieri, S., Turini, F.: Measuring discrimination in socially-sensitive decision records. In: Proceedings of the SIAM Int’l Conf. on Data Mining, pp. 499–514. Citeseer (2009)Google Scholar
- 12.Resnick, P., Konstan, J., Jameson, A.: Measuring discrimination in socially-sensitive decision records. In: Proceedings of the 5th ACM Conference on Recommender Systems: Panel on The Filter Bubble, pp. 499–514 (2011)Google Scholar
- 14.Vapnik, V.N.: Statistical learning theory (1998)Google Scholar
- 15.Zemel, R.S., Wu, Y., Swersky, K., Pitassi, T., Dwork, C.: Learning fair representations. In: ICML (3). JMLR Proceedings, vol. 28, pp. 325–333. JMLR.org (2013)Google Scholar
- 16.Zliobaite, I., Kamiran, F., Calders, T.: Handling conditional discrimination. In: 2011 IEEE 11th International Conference on Data Mining (ICDM), pp. 992–1001. IEEE (2011)Google Scholar