Optimizing Data and Compute for AI-Driven Applications in Resource-Limited Environments
发布者：梁慧丽发布时间：2025-11-28浏览次数：17
Jiacheng Liu
Hong Kong University of Science and Technology
2025年5月27日 星期二
下午 14:00-15:00
102报告厅
While modern AI systems demonstrate remarkable capabilities, their growing data and computational demands hinder widespread adoption across many domains. Over the past several years, my research has focused on making AI more accessible in resource-limited environments through three complementary innovations. My key contributions include: (1) novel data efficiency frameworks for label inference and noisy label learning that enable high-quality model training; (2) advanced compute optimization techniques featuring situation-aware execution mechanisms that dramatically reduce the computational footprint of large language models; and (3) successful real-world applications in biological science and systems engineering. Together, these contributions help close the gap between the promise of state-of-the-art AI and its practical, widespread deployment in domains facing real-world constraints.
Dr. Jiacheng Liu currently serves as a Postdoctoral Fellow at the Hong Kong University of Science and Technology. Prior to this appointment, he was a Postdoctoral Fellow at the Chinese University of Hong Kong. He earned his Ph.D. degree from Shanghai Jiao Tong University. His research interests span artificial intelligence, computing systems, and their real-world applications, with particular focus on developing more accurate and efficient AI solutions. He has published more than 30 papers in top-tier journals and conferences such as AAAI, USENIX ATC, Advanced Science, etc. He is the recipient of 3 Best Paper (Nominee) Awards from CCF recommended conferences. He also received the AAAI/ACM SIGAI New and Future AI Educators Award.
