My work broadly deals with investigating the effects of changing data distributions in Machine Learning, and their implications on Fairness in ML. I am interested in both the theoretical and practical aspects of such phenomena. One of our works investigated the effects of under-representation and labeling biases on some commonly used fair classifiers, and contained some almost obvious theoretical observations. Another work (in review) looks at investigating the phenomena of recovering Bayes optimal classifiers with biased data using fairness constraints, which is an extension of this work by Blum et al. Currently, we are also working on investigating the effects of noisy or uncertain protected attributes on fair classifiers, which started as an internship project with Prof. Shin’ichi Satoh at the National Institute of Informatics at Tokyo, Japan.