The Front-Edge of Decentralized Artificial Intelligence and Machine Learning
- Dr. Shuai Li, University of Cambridge
- Time: 2018-01-12 14:30
- Host: Prof. Baoquan Chen
- Venue: Room 207, Courtyard No.5, Jingyuan
Abstract
I will talk about algorithmic and theoretical aspects of two distributed techniques in machine learning and artificial intelligence based on recent technological advances in clustering bandits. For the first, the assumption is that all the peers are solving the same linear regression based problem, and show that proposed algorithm achieves the optimal asymptotic regret rate of any centralised algorithm that can instantly communicate information between the peers. For the second, there are clustering structure of peers solving the same bandit problem within each cluster, and prove that proposed algorithm discovers these latent clusters, while achieving the optimal asymptotic convergence rate within each one. Through large-scale benchmark experiments on several real-world data sets, I will demonstrate the performance of devised algorithms comparing with state-of-the-art machine learning methods.
Biography
Dr Shuai Li is a researcher at the University of Cambridge. Dr Li has 8+ years professional experience across Europe, North America, Middle East, and Asia Pacific; and 20+ years project experience in Information Science and Technology; also Shuai served as Chief Scientist for some stealth startups. He has been engaged in the analysis of complex, dynamic data at scale, and brought cutting-edge machine learning and artificial intelligence to the heart of industrial-scale big data analytics and data science. As the academic service he has been involved in a number of international prestigious conferences including ICML 2018; NIPS 2017, 2016; UAI 2018, 2017; AISTATS 2018, 2017; WWW 2018, 2016; SDM 2018, 2017, 2016; WSDM 2018; SIGIR 2017; SIGKDD 2017; ICDM 2017; IJCAI 2017, etc.