Xueru Zhang awarded Rackham Predoctoral Fellowship

ECE PhD student Xueru Zhang has been awarded a Rackham Predoctoral Fellowship to support her research on developing artificial intelligence (AI) systems that have positive societal impacts while addressing associated ethical issues. She focuses on privacy invasion and discrimination.

Many machine learning algorithms are trained or developed using private data such as medical records, financial records, and online activities. As a result, there is a high risk that personal data will be leaked, resulting in a potentially significant loss to both individuals and data collectors. To address this issue, Zhang has developed algorithms that improve the methods of preserving individual privacy while accomplishing computational goals.

Zhang has also designed mechanisms that can incentivize individuals to voluntarily secure themselves. For the issue of trading private data, where individuals are compensated for sharing their private data if privacy violations cannot be avoided, Zhang developed a novel algorithm that improves the payment-accuracy tradeoff of the buyer compared to traditional algorithms.

In addition to privacy concerns, machine learning models developed from real-world data can inherit pre-existing bias in their datasets. For example, speech recognition products such as Amazon’s Alexa and Google Home have been shown to struggle with accents, and the COMPAS algorithm used by courts in the United States for recidivism prediction has been shown to be biased against Black defendants. Moreover, bias in the decisions made by such models can be captured in datasets, which are then used to train future models, further perpetuating bias.

Due to these issues, Zhang has designed algorithms to incorporate proper fairness interventions that promote long-term social equality. These interventions help identify and address the human element when designing machine learning algorithms. She plans to continue this work and include embedding long-term decision impact into the very definition of algorithmic fairness.

Zhang earned her MS degree in Electrical and Computer Engineering from University of Michigan in 2016, and B.Eng. degree in Electronic and Information Engineering from Beihang University (BUAA), Beijing, China, in 2015. She is advised by Mingyan Liu, the Peter and Evelyn Fuss Chair of ECE.