Engineering DPhil Student Wins Prototypes for Humanity innovation award

Prototypes for Humanity is a community of leading innovators aiming to create a better future through science and research. Its activities raise awareness of global problems, celebrate solutions, and support action for positive social and environmental impact.

More than 2,700 applications were submitted for the 2024 programme from universities worldwide. 100 research-backed projects, selected for their social and environmental impact innovation, were shortlisted and showcased in Dubai - those demonstrating the highest potential to solve global issues, backed by rigorous academic research.

Xinpeng Hong works alongside Professor Noa Zilberman in the Computing Infrastructure research group, which uses micro-architectures to improve system-scale and application level performance. At Prototypes for Humanity, Xinpeng showcased work in In-Network Machine Learning for Ultra-Low Latency in Time-Sensitive Applications, carried out in a team with DPhil student Changgang Zheng. Machine Learning (ML) is increasingly used in time-sensitive applications such as road safety, cybersecurity and financial trading, however traditional server-based frameworks often struggle with latency (delays) and performance. In-network ML, embedding ML in network devices such as Network Interface Cards (NICs), achieves up to 800 times faster response times and uses 1000 times less power compared to server-based solutions.

Xinpeng says, “I am truly honoured to receive this award, which signifies that in-network machine learning is gradually evolving from an emerging field to one more widely recognized for its advantages. Compared to traditional CPU-based solutions, in-network machine learning offers lower latency, higher throughput, and greater energy efficiency, making it increasingly impactful.

I sincerely thank the Dubai Future Solutions – Prototypes for Humanity program for providing such a fantastic platform to showcase my project, as well as the jury members for their recognition and support. I aim to expand the application of in-network machine learning to more low-latency scenarios, improving people’s lives and enhancing efficiency across industries.”

Director of Prototypes for Humanity, Tadeu Baldani Caravieri, said, “The diversity of applications we received for the 2024 programme, covering all fields of sciences, technology and creative studies, reflects a remarkably exciting global state of innovation”.