Lei Jiao, Assistant Professor of Computer Science at the University of Oregon, will lead a new National Science Foundation (NSF) project titled “Collaborative Research: CNS Core: Small: Edge AI with Streaming Data: Algorithmic Foundations for Online Learning and Control.” This three-year project is in collaboration with Xiaojun Lin, Professor of Electrical and Computer Engineering at the Purdue University, and has a total budget of about $600,000.
Real-time Artificial Intelligence (AI) upon online streaming data is crucial for many emerging applications such as augmented reality and smart healthcare. Edge AI moves AI to the edge computing infrastructures closer to end users and devices where data are generated, reducing communication latency and enabling fast inference decisions. Yet, towards realizing real-time AI, edge AI for online streaming data poses significant challenges due to the unpredictable dynamism of the streaming data and the inherent limitation of the edge computation/communication capability. This project addresses these challenges via three closely-related thrusts: (i) developing online learning algorithms to identify machine learning models of the best inference accuracy for dynamic deployments on edge servers, while accounting for heterogeneous switching and feedback costs; (ii) designing distributed online transfer learning methods to update and retrain machine learning models at edge upon streaming data while adapting to concept drifts; and (iii) devising partial-index-based control policies to optimize the timeliness of interactive edge AI services under tight resource constraints.
Prof. Jiao researches edge computing and cloud computing, spanning distributed AI, security, energy, and multimedia. He has published 55 research papers, with the majority in prestigious or important journals and conferences, which have attracted 4200+ citations (according to Google Scholar). See details on his personal website.