ICS Students Win 2024 Gen AI Hackathon

At the 2024 GenAI Hackathon, the goal was to revolutionize how data is shared within digital ecosystems, using generative AI to develop tools that enhance data privacy and protection. The online event, hosted by UCLA’s Trustworthy AI Lab and the Global Entrepreneurship Society, kicked off on May 29 and ended June 28.

“We were tasked with analyzing some data from two parties, an advertiser and a publisher, to enhance the click-through rate of users to the ads,” says Lucas Chin, a second-year data science major in UC Irvine’s Donald Bren School of Information and Computer Sciences (ICS). “The catch was that both parties had confidential data that wasn’t to be shared, so each team had to create a ‘data clean room’ to allow for access to both parties’ data while maintaining user privacy and data security.”

Chin, working with Chi Lu, also a data science undergraduate at UCI, along with Chloe Houvardas of Queen’s University and Daviya Shah of New York University, created DataGuardians, a virtual machine for securely handling and processing confidential data.

“Through the data clean room, we worked to analyze the data, use machine learning models to predict features, and then synthetically generate a replica dataset to allow for use outside of the data clean room,” explains Chin. In particular, the team built a generative adversarial network to maintain privacy.

The team’s efforts paid off, as they took first place in the hackathon, winning $700 and gaining practical experience in how to enhance data privacy.

“The task presented mimicked industry needs,” says Chin, “so it was a good introduction to real-world scenarios regarding generative AI and data security.”

— Shani Murray