<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
</head>
<body>
<div class="BodyFragment"><font size="2"><span style="font-size:11pt;">
<div class="PlainText">This is an announcement of Emily Wenger's Candidacy Exam.<br>
===============================================<br>
Candidate: Emily Wenger<br>
<br>
Date: Friday, June 17, 2022<br>
<br>
Time: 10:30 am CST<br>
<br>
Remote Location: <a href="https://uchicago.zoom.us/j/4571855061?pwd=d1BURDhucEVOR1JjaVY2V2cvRENWZz09">
https://uchicago.zoom.us/j/4571855061?pwd=d1BURDhucEVOR1JjaVY2V2cvRENWZz09</a>; Meeting ID: 457 185 5061, Passcode: 645188<br>
<br>
Location: JCL 298<br>
<br>
Title: Towards Reclaiming Data Agency in the Age of Ubiquitous Machine Learning<br>
<br>
Abstract: As machine learning (ML) models have expanded in size and scope in recent years, so has the amount of data needed to train them. This creates privacy risks for individuals whose data -- be it their images, emails, tweets, or browsing history -- is
used for training. For example, ML models can memorize their training data, revealing private information about individuals in the dataset. Furthermore, users whose data is co-opted for ML use may end up enrolled in a privacy-compromising system, such as a
large-scale facial recognition model. Most existing work on ML data privacy accepts the premise that data use is inevitable and instead tries to mitigate privacy risks during model training. However, privacy-conscious individuals may desire agency over how
and if their data is used, rather than only having their privacy preserved when it is used. Data agency, the ability to know and control how and if one's data is used in ML systems, is an important complement to existing privacy protection approaches, and
it is the focus of this thesis. <br>
Data agency can take many forms, and this thesis will develop technical solutions that enable individuals to disrupt or discover when their data is used in large-scale ML systems. It targets data agency in the context of large-scale facial recognition (FR)
systems, providing ways for users to combat unwanted facial recognition. This work proposes three data agency solutions to disrupt or trace data use in FR systems or, in extreme cases, directly attack the FR system. Additionally, it develops a framework for
reasoning about broadly about data agency in the context of FR. It will use this framework to outline both technical and social challenges of data agency solutions in the FR space and propose directions for future research. Finally, this thesis will discuss
the connections between the proposed FR-specific data agency solutions and methods for reclaiming data agency in other domains.<br>
<br>
Advisors: Ben Zhao and Heather Zheng<br>
<br>
Committee Members: Ben Zhao, Heather Zheng, Yuxin Chen, and Aloni Cohen<br>
<br>
</div>
</span></font></div>
<div class="BodyFragment"><font size="2"><span style="font-size:11pt;">
<div class="PlainText"><br>
<br>
</div>
</span></font></div>
</body>
</html>