Steven He, a data science student from the Class of 2024 at Duke Kunshan University, was the only undergraduate student to present his research at the Flower AI Summit 2024 in London, the world’s largest federated learning conference.
The Flower AI Summit, initiated in 2021 by the Flower team at Cambridge University, annually assembles the brightest minds in federated learning—a field at the forefront of privacy-conscious, collaborative machine learning. It has brought together leading scientists from institutions and corporations like Stanford, Carnegie Mellon, Cambridge, Oxford, Google DeepMind, Amazon Web Services, Samsung, and Volvo. Proceedings are broadcast to a global audience live.
Under the mentorship of Dr. Bing Luo, assistant professor of data and co-presented FedCampus, a trailblazing health data analysis initiative based on federated learning and optimized for next-gen telecom networks. This student privacy-centric project performs intelligent analytics on metrics such as sleep quality, heart rate, and stress levels, aiding campus administrators in health trend monitoring.
He’s innovative solution, FedKit, introduced on the open-source platform Flower, facilitates model conversions and accelerates cross-platform training. He developed FedKit after joining Luo’s research team through Duke Kunshan’s Summer Research Scholars program in 2023, when he had difficulty reconciling federated learning applications across Android and iOS devices.
Reflecting on his research experience, He noted, “Engaging with my peers in federated learning worldwide was eye-opening, particularly seeing the breakthroughs and diverse applications extending beyond traditional tech disciplines.”
He’s work prompted immediate applications, such as the deployment of his framework by a Slovenian university and enhancements by a South Korean team led by Farkhod Yusubov to their mobile “FedOps” provider.
He added, “The university’s rich resources profoundly enabled my exploration in the AI field.”
He’s research achievements also include a system demonstration paper—”FedKit: Enabling Cross-Platform Federated Learning for Android and iOS,”—accepted by IEEE INFOCOM 2024 this February.
As the first author, He collaborated with Luo, Beilong Tang (’25), Boyan Zhang and Renyuan Zhang (’26), Luyao Wang (’24), and postgraduate student Qingning Zeng from the Electrical and Computer Engineering program at Duke Kunshan, as well as peers from the Hong Kong University of Science and Technology, the University of California at Los Angeles, and the Flower Lab.
Luo lauds this achievement as a testament to He’s talents and hard work and Duke Kunshan’s vibrant research culture.
He will soon pursue his Ph.D. studies in Computer Science at the University of Southern California on full funding.
