----------------------------------------------------------------------------------------------------







       Department of Systems Engineering and Engineering Management



                          The Chinese University of Hong Kong



----------------------------------------------------------------------------------------------------



Time: 9:00-10:15, 9th May (Online, Zoom)



Zoom Meeting Info:



************************************************************************************



Topic: CUHK-SEEM Seminar by Bradley Sturt

Time: May 9, 2025 09:00 AM Hong Kong SAR

Join Zoom Meeting

https://cuhk.zoom.us/j/4084657934?pwd=U7JSn4uZKUQz9bjdNSzLWayX5JdNgT.1&omn=91646888436



Meeting ID: 408 465 7934

Passcode: cuhk



************************************************************************************



Title: Improving the security of United States elections with robust 

optimization



Speaker: Professor Brad Sturt, University of Illinois Chicago



Abstract: For more than a century election officials across the United 

States have inspected voting machines before elections using a 

procedure called Logic and Accuracy Testing (LAT). This procedure 

consists of election officials casting a test deck of ballots into 

each voting machine and confirming the machine produces the expected 

vote total for each candidate. In this talk I will bring a scientific 

perspective to LAT by introducing the first formal approach to 

designing test decks with rigorous security guarantees. Specifically 

we propose using robust optimization to find test decks that are 

guaranteed to detect any voting machine misconfiguration that would 

cause votes to be swapped across candidates. Out of all the test decks 

with this security guarantee the robust optimization problem yields 

the test deck with the minimum number of ballots thereby minimizing 

implementation costs for election officials. To facilitate deployment 

at scale we developed a practical exact algorithm for solving our 

robust optimization problems based on mixed-integer optimization and 

the cutting plane method. In partnership with the Michigan Bureau of 

Elections we retrospectively applied our robust optimization approach 

to all 6928 ballot styles from Michigan's November 2022 general 

election. This retrospective study reveals that the test decks with 

rigorous security guarantees obtained by our approach require on 

average only 1.2% more ballots than current practice. Our robust 

optimization approach has since been piloted in real-world elections 

by the Michigan Bureau of Elections as a low-cost way to improve 

election security and increase public trust in democratic institutions.



Short Bio: Brad Sturt is an assistant professor of business analytics 

at the University of Illinois Chicago. His research interest is 

optimization under uncertainty with focus on applications in 

operations revenue management and the public sector. Recent 

applications have included election administration, data-driven 

assortment planning and pricing, and high-dimensional optimal 

stopping. His research has received several recognitions including 

second place in the INFORMS Junior Faculty Interest Group (JFIG) Paper 

Competition and second place in the INFORMS George Nicholson Student 

Paper Competition. Outside of academia he is a co-founder of BallotIQ, 

an election administration startup.